Upload folder using huggingface_hub

.dockerignore

@@ -0,0 +1,11 @@
+.git
+__pycache__
+*.pyc
+.venv
+venv
+.env
+node_modules
+results
+data
+*.ckpt
+*.h5

.gitattributes

@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+sample_files/Screenshot[[:space:]]2024-07-13[[:space:]]163331.png filter=lfs diff=lfs merge=lfs -text
+sample_files/alperen_celik_14_08.pdf filter=lfs diff=lfs merge=lfs -text
+sample_files/medium_article_image.jpg filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -0,0 +1,28 @@
+FROM python:3.10-slim
+
+# Install system packages required for OCR (Tesseract, poppler for PDF tools)
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+       tesseract-ocr \
+       libtesseract-dev \
+       libleptonica-dev \
+       pkg-config \
+       poppler-utils \
+       build-essential \
+       git \
+    && rm -rf /var/lib/apt/lists/*
+
+# Copy and install Python dependencies
+COPY requirements.txt /tmp/requirements.txt
+RUN python -m pip install --upgrade pip && \
+    pip install --no-cache-dir -r /tmp/requirements.txt
+
+# Copy application
+COPY . /app
+WORKDIR /app
+
+# Expose default port (Spaces will set PORT env var)
+ENV PORT=7860
+
+# Run Streamlit app on container start
+CMD bash -lc "streamlit run streamlit_app.py --server.port ${PORT} --server.address 0.0.0.0"

README.md

@@ -1,10 +1,226 @@
+# OCRInsight
+
 ---
-title: Ocr
-emoji: ⚡
-colorFrom: blue
-colorTo: green
+title: OCRInsight
+emoji: "🧾"
+colorFrom: "0A84FF"
+colorTo: "7C3AED"
 sdk: docker
+sdk_version: "1.0"
+app_file: streamlit_app.py
 pinned: false
 ---
+This Streamlit application allows users to perform OCR (Optical Character Recognition) using multiple open-source OCR engines and optionally process the OCR results using LLMs (Large Language Models). Users can compare the outputs of different OCR models and perform tasks such as summarization or text generation based on the OCR results.
+
+Here the link of comprehensive explanation: (https://medium.com/@alperenclk/ocrinsight-building-a-modular-ocr-and-llm-application-f3d3a1ea7a18)
+![alt text](https://github.com/Alperenclk/All_OCR-s_tools/blob/main/sample_files/sample_screen.png?raw=true)
+
+## Features
+### Multiple OCR Engines Supported:
+
+* EasyOCR
+* DocTR
+* Tesseract OCR
+* PaddleOCR
+
+#### Optional LLM Processing:
+
+Use models like llama3.1, llama3, gemma2 via Ollama.
+Perform tasks such as summarization or text generation based on OCR results.
+
+#### Compare OCR Outputs:
+
+Select multiple OCR models to compare their outputs side by side.
+#### Save Outputs:
+
+Option to save OCR and LLM outputs to text files.
+
+## Installation
+### Prerequisites
+- Python 3.7 or higher
+- pip package manager
+
+### Clone the Repository
+
+```bash 
+git clone https://github.com/Alperenclk/OCRInsight-open-source-OCRs-Plus-LLM.git
+cd ocr-llm-app
+```
+
+### Create a Virtual Environment (Recommended)
+```bash 
+python -m venv venv
+source venv/bin/activate  # On Windows use: venv\Scripts\activate
+```
+
+### Install Required Python Packages
+#### Install the required packages using pip:
+
+```bash 
+pip install -r requirements.txt
+```
+Note: The requirements.txt file includes basic dependencies. Depending on the OCR engines and LLM support you want to use, you may need to install additional dependencies as described below.
+
+## Install OCR Engine Dependencies
+
+### EasyOCR
+```bash 
+pip install easyocr
+```
+
+### DocTR
+```bash 
+pip install python-doctr[torch]
+```
+Note: For GPU support, ensure that PyTorch is installed with CUDA support.
+
+### Tesseract OCR
+Install Tesseract OCR Engine:
+
+#### Windows:
+
+Download the Tesseract installer from UB Mannheim: <https://github.com/UB-Mannheim/tesseract/wiki>.
+
+**Run the installer and follow the instructions.
+Note the installation path (e.g., C:\Program Files\Tesseract-OCR\tesseract.exe).
+Update the pytesseract.pytesseract.tesseract_cmd variable in ocr_engines.py to point to the Tesseract executable.**
+
+#### macOS:
+
+```bash 
+brew install tesseract
+```
+#### Ubuntu/Linux:
+
+``` bash 
+sudo apt-get update
+sudo apt-get install tesseract-ocr
+```
+
+##### Install Python Wrapper:
+
+```bash 
+pip install pytesseract
+```
+##### Language Data Files:
+
+Ensure that the language data files for the languages you intend to use are installed. For example, to install Turkish language data on Ubuntu:
+
+```bash 
+sudo apt-get install tesseract-ocr-tur
+```
+
+### PaddleOCR
+#### Install PaddlePaddle:
+
+#### CPU Version:
+
+```bash 
+pip install paddlepaddle
+```
+#### GPU Version:
+
+Refer to the PaddlePaddle Installation Guide for GPU support.
+
+### Install PaddleOCR:
+
+```bash 
+pip install paddleocr
+```
+
+## Install LLM Dependencies (Optional)
+If you want to use the LLM features, install **Ollama**:
+
+```bash
+pip install ollama
+```
+Note: If you do not wish to use the LLM features, **you can skip this step**. The application will work in OCR-only mode.
+
+## Usage
+### Run the Application
+```bash 
+streamlit run app.py
+```
+
+## Application Interface
+### Settings Sidebar:
+
+**Select Device:** Choose between CPU and GPU (if available).
+
+**Language Selection:** Choose the language for OCR processing.
+
+**Select OCR Models:** Choose one or more OCR models to use.
+
+**LLM Model Selection:** Choose an LLM model or select "Only OCR Mode" to disable LLM features.
+
+**LLM Command and Task Type:** Enter commands and select tasks if LLM is enabled.
+
+**Save Outputs:** Option to save OCR and LLM outputs to files.
+
+### Main Area:
+
+**File Upload:** Upload a PDF or image file for OCR processing.
+
+**OCR Results:** View the OCR results from the selected models.
+
+**LLM Processing:** Perform LLM processing on the combined OCR text (if enabled).
+
+## Notes
+**Language Support:**
+
+Ensure that the necessary language data files or models are installed for each OCR engine you intend to use.
+Some OCR engines may require specific language codes or configurations.
+
+**GPU Support:**
+
+For GPU acceleration, ensure that your hardware supports it and that the necessary libraries (e.g., CUDA) are installed.
+Not all OCR engines support GPU acceleration.
+
+**Performance:**
+
+Processing multiple OCR engines simultaneously may consume significant resources.
+Processing large files or images may take longer.
+Modular Code Structure
+The application is structured modularly to enhance maintainability and extensibility.
+
+**app.py:** The main Streamlit application script.
+
+**ocr_engines.py:** Contains functions to initialize and perform OCR using different engines.
+
+**llm_processor.py:** Contains functions for LLM processing (optional).
+Modifying the Code
+
+#### **Adding a New OCR Engine:**
+
+Create a new function in ocr_engines.py to initialize and perform OCR with the new engine.
+Update initialize_ocr_models and perform_ocr functions accordingly.
+
+**Modifying LLM Functionality:**
+
+Update llm_processor.py with new LLM models or processing methods.
+
+**Disabling LLM Features:**
+
+If you don't want to use LLM features, you don't need to install ollama.
+The application will automatically disable LLM features if ollama is not installed.
+
+## Troubleshooting
+**Import Errors:**
+
+If you encounter import errors, ensure that all required packages are installed.
+For optional features (like LLM), missing packages will disable those features without affecting the rest of the application.
+
+**Tesseract Not Found:**
+
+Ensure that the Tesseract executable path is correctly set in ocr_engines.py.
+Verify that Tesseract is installed and the path is correct.
+
+**Language Data Missing:**
+
+Install the necessary language data files for the OCR engines.
+Contributing
+Contributions are welcome! Please fork the repository and submit a pull request for any improvements or new features.
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+### License
+This project is licensed under the **MIT** License.
+# OCR

README_DEPLOY.md

@@ -0,0 +1,30 @@
+Deployment notes for Hugging Face Spaces
+
+1) HF_TOKEN secret
+- Create a Hugging Face token at https://huggingface.co/settings/tokens
+- Token should have repository write permissions (to create and push Spaces)
+- In GitHub, go to Settings -> Secrets -> Actions -> New repository secret
+  - Name: HF_TOKEN
+  - Value: <your_token_here>
+
+2) Streamlit compatibility
+- The workflow creates the Space with `space_sdk='streamlit'` so it will run as a Streamlit app.
+- Hugging Face Spaces will run `streamlit_app.py` or `app.py` by default; this repo contains `streamlit_app.py` to be explicit.
+
+3) System dependencies
+- Some OCR engines require system packages (e.g., Tesseract binary, system libs for PaddlePaddle). Hugging Face's Streamlit SDK does not allow installing system packages.
+- If you need system packages, use a Docker-based Space (set `space_sdk='docker'` and add a Dockerfile that installs required system packages).
+
+4) LLM / Ollama
+- The app optionally uses `ollama` for LLM features. Ollama is not installed by default in Spaces; LLM features will be disabled if `ollama` isn't present.
+
+5) Tesseract
+- Ensure Tesseract is available in the environment or use the Docker approach to install it.
+
+6) Running CI/CD
+- After pushing to `main` and setting `HF_TOKEN` secret, the GitHub Actions workflow `.github/workflows/deploy_to_hf.yml` will create the Space and upload the repository.
+
+Note: This repository includes a `Dockerfile` and the CI workflow is configured to create a Docker-based Space (`space_sdk='docker'`). The Dockerfile installs system dependencies such as Tesseract so the OCR engines can run inside the Space container.
+
+7) Troubleshooting
+- If the deployment fails, open the Actions run logs to see the error and adjust the workflow or repository accordingly.

app.py

@@ -0,0 +1,197 @@
+import streamlit as st
+from PIL import Image
+import fitz  # PyMuPDF
+import numpy as np
+import tempfile
+import os
+import time
+import io
+import json
+import torch
+import cv2
+
+# Import OCR engines
+import ocr_engines
+
+# Try importing LLM processor if LLM features are to be used
+llm_available = False
+try:
+    import llm_processor
+
+    llm_available = True
+except ImportError:
+    pass  # LLM features will be disabled
+
+# Create results folder if it doesn't exist
+if not os.path.exists("results"):
+    os.makedirs("results")
+
+# Streamlit application
+st.title("OCRInsight")
+
+# Sidebar
+st.sidebar.header("Settings")
+
+
+# Function to save text to file
+def save_text_to_file(attributes_of_output, all_ocr_text, filename):
+    with open(filename, "a", encoding="utf-8") as f:
+        f.write("\n" + "-" * 75 + "\n")
+        f.write("Attributes of Output:\n")
+        f.write(attributes_of_output)
+        f.write("\nOCR Result:\n")
+        f.write(all_ocr_text)
+        f.write("\n" + "-" * 75 + "\n")
+    st.success(f"{filename} saved successfully!")
+
+
+# Device selection
+device = st.sidebar.radio("Select Device", ["CPU", "GPU (CUDA)"])
+save_output = st.sidebar.checkbox("Save Outputs")
+
+# Language selection
+language = st.sidebar.selectbox(
+    "Select Language", ["Türkçe", "English", "Français", "Deutsch", "Español"]
+)
+
+# Map selected language to language codes
+language_codes = {
+    "Türkçe": "tr",
+    "English": "en",
+    "Français": "fr",
+    "Deutsch": "de",
+    "Español": "es",
+}
+
+# OCR model selection
+ocr_models = st.sidebar.multiselect(
+    "Select OCR Models",
+    ["EasyOCR", "DocTR", "Tesseract", "PaddleOCR"],
+    ["EasyOCR"],  # default selection
+)
+
+# LLM model selection
+llm_model = st.sidebar.selectbox(
+    "Select LLM Model", ["Only OCR Mode", "llama3.1", "llama3", "gemma2"]
+)
+
+# Conditional UI elements based on LLM model selection
+if llm_model != "Only OCR Mode" and llm_available:
+    user_command = st.sidebar.text_input("Enter command:", "")
+
+    task_type = st.sidebar.radio("Select task type:", ["Summarize", "Generate"])
+elif llm_model != "Only OCR Mode" and not llm_available:
+    st.sidebar.warning(
+        "LLM features are not available. Please install 'ollama' to enable LLM processing."
+    )
+    llm_model = "Only OCR Mode"
+
+# Check GPU availability
+if device == "GPU (CUDA)" and not torch.cuda.is_available():
+    st.sidebar.warning("GPU (CUDA) not available. Switching to CPU.")
+    device = "CPU"
+
+# Initialize OCR models
+ocr_readers = ocr_engines.initialize_ocr_models(
+    ocr_models, language_codes[language], device
+)
+
+# File upload
+uploaded_file = st.file_uploader(
+    "Upload File (PDF, Image)", type=["pdf", "png", "jpg", "jpeg"]
+)
+
+# Create results folder if it doesn't exist
+if not os.path.exists("results"):
+    os.makedirs("results")
+
+if uploaded_file is not None:
+    start_time = time.time()
+
+    if uploaded_file.type == "application/pdf":
+        pdf_document = fitz.open(stream=uploaded_file.read(), filetype="pdf")
+        images = []
+        for page_num in range(len(pdf_document)):
+            page = pdf_document.load_page(page_num)
+            pix = page.get_pixmap()
+            img_data = pix.tobytes("png")
+            img = Image.open(io.BytesIO(img_data))
+            images.append(img)
+        total_pages = len(pdf_document)
+        pdf_document.close()
+    else:
+        images = [Image.open(uploaded_file)]
+        total_pages = 1
+
+    all_ocr_texts = {
+        model_name: "" for model_name in ocr_models
+    }  # To store OCR text for each model
+
+    for page_num, image in enumerate(images, start=1):
+        st.image(image, caption=f"Page {page_num}/{total_pages}", use_column_width=True)
+
+        # Perform OCR with each selected model
+        for model_name in ocr_models:
+            text = ocr_engines.perform_ocr(
+                model_name, ocr_readers, image, language_codes[language]
+            )
+            all_ocr_texts[
+                model_name
+            ] += f"--- Page {page_num} ({model_name}) ---\n{text}\n\n"
+
+            st.subheader(f"OCR Result ({model_name}) - Page {page_num}/{total_pages}:")
+            st.text(text)
+
+    end_time = time.time()
+    process_time = end_time - start_time
+
+    st.info(f"Processing time: {process_time:.2f} seconds")
+
+    # Save OCR outputs if selected
+    if save_output:
+        attributes_of_output = {
+            "Model Names": ocr_models,
+            "Language": language,
+            "Device": device,
+            "Process Time": process_time,
+        }
+        for model_name, ocr_text in all_ocr_texts.items():
+            filename = f"results//ocr_output_{model_name}.txt"
+            save_text_to_file(
+                json.dumps(attributes_of_output, ensure_ascii=False), ocr_text, filename
+            )
+
+    # LLM processing
+    if (
+        llm_model != "Only OCR Mode"
+        and llm_available
+        and st.sidebar.button("Start LLM Processing")
+    ):
+        st.subheader("LLM Processing Result:")
+
+        # Combine all OCR texts
+        combined_ocr_text = "\n".join(all_ocr_texts.values())
+
+        # Prepare the prompt based on the task type
+        if task_type == "Summarize":
+            prompt = f"Please summarize the following text. Command: {user_command}\n\nText: {combined_ocr_text}"
+        else:  # "Generate"
+            prompt = f"Please generate new text based on the following text. Command: {user_command}\n\nText: {combined_ocr_text}"
+
+        llm_output = llm_processor.process_with_llm(llm_model, prompt)
+
+        # Display the result
+        st.write(f"Processing completed using '{llm_model}' model.")
+        st.text_area("LLM Output:", value=llm_output, height=300)
+
+        # Save LLM output if selected
+        if save_output:
+            filename = "llm_output.txt"
+            save_text_to_file(llm_output, "", filename)
+
+elif llm_model != "Only OCR Mode" and not llm_available:
+    st.warning(
+        "LLM features are not available. Please install 'ollama' to enable LLM processing."
+    )
+
+st.sidebar.info(f"Selected device: {device}")

llm_processor.py

@@ -0,0 +1,17 @@
+# llm_processor.py
+
+import ollama
+
+
+def process_with_llm(llm_model, prompt):
+    response = ollama.chat(
+        model=llm_model,
+        messages=[
+            {
+                "role": "user",
+                "content": prompt,
+            },
+        ],
+    )
+    llm_output = response["message"]["content"]
+    return llm_output

ocr_engines.py

@@ -0,0 +1,137 @@
+"""OCR engine initializers and runners with safer Tesseract handling."""
+
+import os
+import sys
+import tempfile
+import numpy as np
+
+try:
+    import easyocr
+except Exception:
+    easyocr = None
+
+try:
+    from doctr.io import DocumentFile
+    from doctr.models import ocr_predictor
+except Exception:
+    DocumentFile = None
+    ocr_predictor = None
+
+try:
+    from paddleocr import PaddleOCR
+except Exception:
+    PaddleOCR = None
+
+try:
+    import pytesseract
+except Exception:
+    pytesseract = None
+
+try:
+    import cv2
+except Exception:
+    cv2 = None
+
+
+def initialize_ocr_models(ocr_models, language_code, device):
+    ocr_readers = {}
+
+    if "EasyOCR" in ocr_models and easyocr is not None:
+        ocr_readers["EasyOCR"] = easyocr.Reader(
+            [language_code], gpu=(device == "GPU (CUDA)")
+        )
+
+    if "DocTR" in ocr_models and ocr_predictor is not None:
+        ocr_readers["DocTR"] = ocr_predictor(pretrained=True)
+
+    if "PaddleOCR" in ocr_models and PaddleOCR is not None:
+        use_gpu = True if device == "GPU (CUDA)" else False
+        ocr_readers["PaddleOCR"] = PaddleOCR(lang=language_code, use_gpu=use_gpu)
+
+    # Tesseract: only set executable path for known Windows locations; on Unix, assume tesseract is on PATH
+    if "Tesseract" in ocr_models and pytesseract is not None:
+        if sys.platform.startswith("win"):
+            # common Windows installation path
+            pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"
+        else:
+            # check common unix paths and set if tesseract binary exists there
+            for p in ("/usr/bin/tesseract", "/usr/local/bin/tesseract"):
+                if os.path.exists(p):
+                    pytesseract.pytesseract.tesseract_cmd = p
+                    break
+
+    return ocr_readers
+
+
+def perform_ocr(model_name, ocr_readers, image, language_code):
+    text = ""
+
+    if model_name == "EasyOCR":
+        reader = ocr_readers.get("EasyOCR")
+        if reader is None:
+            return "[EasyOCR not available]"
+        result = reader.readtext(np.array(image))
+        text = "\n".join([res[1] for res in result])
+
+    elif model_name == "DocTR":
+        predictor = ocr_readers.get("DocTR")
+        if predictor is None or DocumentFile is None:
+            return "[DocTR not available]"
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as tmp_file:
+            image.save(tmp_file, format="PNG")
+        file_path = tmp_file.name
+        doc = DocumentFile.from_images(file_path)
+        result = predictor(doc)
+        # Safely iterate pages/blocks
+        pages = []
+        for page in result.pages:
+            page_text_blocks = []
+            for block in page.blocks:
+                lines = [" ".join([word.value for word in line.words]) for line in block.lines]
+                page_text_blocks.append("\n".join(lines))
+            pages.append("\n\n".join(page_text_blocks))
+        text = "\n\n".join(pages)
+        try:
+            os.unlink(file_path)
+        except Exception:
+            pass
+
+    elif model_name == "PaddleOCR":
+        reader = ocr_readers.get("PaddleOCR")
+        if reader is None:
+            return "[PaddleOCR not available]"
+        result = reader.ocr(np.array(image))
+        # result may be empty or structured per line
+        try:
+            text = "\n".join([line[1][0] for line in result[0]])
+        except Exception:
+            # fallback: join any text tokens found
+            tokens = []
+            for page in result:
+                for line in page:
+                    if len(line) > 1 and isinstance(line[1], (list, tuple)):
+                        tokens.append(line[1][0])
+            text = "\n".join(tokens)
+
+    elif model_name == "Tesseract":
+        if pytesseract is None:
+            return "[pytesseract not available]"
+        # Convert PIL image to RGB if not already
+        try:
+            if image.mode != "RGB":
+                image = image.convert("RGB")
+        except Exception:
+            pass
+        # Convert image to OpenCV format if cv2 is available
+        if cv2 is not None:
+            opencv_image = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
+        else:
+            # fallback: use raw numpy array
+            opencv_image = np.array(image)
+        config = f"--oem 3 --psm 6 -l {language_code}"
+        try:
+            text = pytesseract.image_to_string(opencv_image)  # , config=config
+        except Exception as e:
+            text = f"[Tesseract error: {e}]"
+
+    return text

requirement.txt

@@ -0,0 +1,11 @@
+streamlit
+Pillow
+PyMuPDF
+numpy
+torch
+easyocr
+python-doctr[torch]
+paddlepaddle  # For CPU; for GPU, specify the appropriate version
+paddleocr
+pytesseract
+opencv-python

requirements.txt

@@ -0,0 +1,11 @@
+streamlit
+Pillow
+PyMuPDF
+numpy
+torch
+easyocr
+python-doctr[torch]
+paddlepaddle
+paddleocr
+pytesseract
+opencv-python

results/ocr_output_DocTR.txt

@@ -0,0 +1,572 @@
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 1.2372400760650635}
+OCR Result:
+--- Page 1 (DocTR) ---
+Genel olarak, sag eliniz ne kadar yi çalisti?
+Sag parmaklanniz ne kadar iyi hareket etti?
+Sag bileginiz ne kadar yi hareket etti?
+Sag elinizin kuvveti nasildi?
+Sag elinizde duyu (his) nasildi?
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR", "EasyOCR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 1.7959399223327637}
+OCR Result:
+--- Page 1 (DocTR) ---
+Genel olarak, sag eliniz ne kadar yi çalisti?
+Sag parmaklanniz ne kadar iyi hareket etti?
+Sag bileginiz ne kadar yi hareket etti?
+Sag elinizin kuvveti nasildi?
+Sag elinizde duyu (his) nasildi?
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR", "EasyOCR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 1.8714756965637207}
+OCR Result:
+--- Page 1 (DocTR) ---
+Genel olarak, sag eliniz ne kadar yi çalisti?
+Sag parmaklanniz ne kadar iyi hareket etti?
+Sag bileginiz ne kadar yi hareket etti?
+Sag elinizin kuvveti nasildi?
+Sag elinizde duyu (his) nasildi?
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 1.326887845993042}
+OCR Result:
+--- Page 1 (DocTR) ---
+Genel olarak, sag eliniz ne kadar yi çalisti?
+Sag parmaklanniz ne kadar iyi hareket etti?
+Sag bileginiz ne kadar yi hareket etti?
+Sag elinizin kuvveti nasildi?
+Sag elinizde duyu (his) nasildi?
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 9.680048942565918}
+OCR Result:
+--- Page 1 (DocTR) ---
+ALPEREN ÇELIK
+- +90 5453851876 peradkisfotgmalcom im.com/m/Aperenell-791aits, 0 llh-com/Alperencls,
+Education
+Afyon Kocatepe University
+Sep. 2018 - Jan 2024
+Bachelor of Mechatronic Engineering
+3.1 gpa
+Relevant Coursework
+Artificial Intelligence
+Software Methodology
+Database Management
+Internet Technology
+Computer Vision
+Algorithms Analysis
+Data Structures
+Systems Programming
+Experience
+Novelty AI
+Sep 2023 - Present
+AI/ML Engineer
+Gebze, Turkiye
+I developed dentification system software for a bank using deep learning techniques. This system aimed to increase
+security by optimizing customer authentication processes and was successfully implemented.
+At a defense industry company, I managed the installation of industrial robots (ABB) on the production line. In this
+project, I provided complex automation solutions to integrate robotic: systems and increase operational efficiency-
+For one of the leading telecom companies in Turkey, I developed software that enables live broadcasting and OTT
+automation with Suitest software using Python and image processing techniques.
+For a beverage company, I was part of the team that developed an artificial intelligence application that checks the
+recognition and accuracy of product labels on the production line.
+University of Malta
+Jul 2023 - Aug 2023 (3 mos)
+AI Researcher
+Maita
+* I worked on a sem-autonomous drone that tries to detect. waste pet bottles on beaches with artificial intelligence. I used
+Python and C++ languages in this project
+TC Diyanet isleri Bagkanhg
+May 2019 - Jul 2022 (3 yrs 3 mos)
+Civil Servant
+Ayonkarhisar, Turkiye
+While studying Mechatronics Engineering at the university, I worked as a civil servant and eammed a living. During these
+three years, the most essential value that this job added to me was to develop myself discipline and determination in
+order to keep the tough school and work life: in balance.
+DHMI Erzurum Airport
+Jul 2022 - Aug 2022 (2 mos)
+Mechatronics Engineer Intern
+Brzurum, Turkiye
+* I worked as an intern in areas such as sensors, electronic cards, x-ray devices in terminal electronics. My most significant
+gain from this internship was learning corporate work discipline and internal relationship techniques.
+Ecodation
+Jun 2021 - Jul 2021 (2 mos)
+Python Developer Intern
+Istanbul, Turkiye
+I made projects such as navigation and customer tracking system for cargo delivery. My main achievement was learning
+to work as a team.
+Projects
+Personalized Product Analysis with AI Python, Net, Huatei Cloud
+Oct 2023
+Our application is designed to help users make informed and healthy choices when purchasing products. By uploading a
+photo of the product's ingredients, the user can get a detailed analysis of how suitable and beneficial the product is for
+them. The application scans the content of the product with artificial intelligence systems and identifies substances that
+may cause allergies or adverse effects that the user has previously dentified. It provides the user with a summary of the
+product's content and the presence or absence of the substances they have identified. Thanks to this application, we
+came 3rd in BIk Academy and Huawei Coding Marathon
+Multi View Breast Cancer Classification App - Python, PyQs, Deep learming
+Apr 2022
+Within the scope of Teknofest artificial intelligence in health competition, the team I captained by developing a
+multi-model deep learning network for the diagnosis of breast cancers succeeded in becoming a finalist.
+Optical Character Recognition with Streamlit I Python, Streamlit, Huggingface
+Jan 2024
+OCR (Optical Character Recognition) technology has transformed how we interact with textual content in the digital
+realm. By converting images, scanned documents, and other media into editable and searchable text, OCR enables us to
+extract valuable information from diverse sources.
+
+--- Page 2 (DocTR) ---
+Technical Skills
+Languages: Python, C/ C++, Matlab, SQL, RobotStudio, Ros, PLC
+Developer Tools: Tensorflow, Pytorch, Google Cloud Platform, Huawei Cloud
+Technologies/Frameworks: Linux, GitHub, Selenium, Docker
+Certificates
+Teknofest Finalist Certificate:13 Foundation
+BTK Academy And Huawei Coding Marathon Certificate of Competitio Winning (3rd)
+EITCA Artificial Intelligence Academy 12 Certificates European Union
+TensorFlow: Advanced Techniques Specialization Coursera
+Google Cloud Expertise Google
+AI Expert Training Program 6 months Republic Of Tirkiye Ministry of Industry and Technology
+Introduction to Machine Learning in Production Coursera
+Hands-on ROS Training with Python :Udemy
+Image processing with deep learning :Udemy
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 9.149980068206787}
+OCR Result:
+--- Page 1 (DocTR) ---
+ALPEREN ÇELIK
+- +90 5453851876 peradkisfotgmalcom im.com/m/Aperenell-791aits, 0 llh-com/Alperencls,
+Education
+Afyon Kocatepe University
+Sep. 2018 - Jan 2024
+Bachelor of Mechatronic Engineering
+3.1 gpa
+Relevant Coursework
+Artificial Intelligence
+Software Methodology
+Database Management
+Internet Technology
+Computer Vision
+Algorithms Analysis
+Data Structures
+Systems Programming
+Experience
+Novelty AI
+Sep 2023 - Present
+AI/ML Engineer
+Gebze, Turkiye
+I developed dentification system software for a bank using deep learning techniques. This system aimed to increase
+security by optimizing customer authentication processes and was successfully implemented.
+At a defense industry company, I managed the installation of industrial robots (ABB) on the production line. In this
+project, I provided complex automation solutions to integrate robotic: systems and increase operational efficiency-
+For one of the leading telecom companies in Turkey, I developed software that enables live broadcasting and OTT
+automation with Suitest software using Python and image processing techniques.
+For a beverage company, I was part of the team that developed an artificial intelligence application that checks the
+recognition and accuracy of product labels on the production line.
+University of Malta
+Jul 2023 - Aug 2023 (3 mos)
+AI Researcher
+Maita
+* I worked on a sem-autonomous drone that tries to detect. waste pet bottles on beaches with artificial intelligence. I used
+Python and C++ languages in this project
+TC Diyanet isleri Bagkanhg
+May 2019 - Jul 2022 (3 yrs 3 mos)
+Civil Servant
+Ayonkarhisar, Turkiye
+While studying Mechatronics Engineering at the university, I worked as a civil servant and eammed a living. During these
+three years, the most essential value that this job added to me was to develop myself discipline and determination in
+order to keep the tough school and work life: in balance.
+DHMI Erzurum Airport
+Jul 2022 - Aug 2022 (2 mos)
+Mechatronics Engineer Intern
+Brzurum, Turkiye
+* I worked as an intern in areas such as sensors, electronic cards, x-ray devices in terminal electronics. My most significant
+gain from this internship was learning corporate work discipline and internal relationship techniques.
+Ecodation
+Jun 2021 - Jul 2021 (2 mos)
+Python Developer Intern
+Istanbul, Turkiye
+I made projects such as navigation and customer tracking system for cargo delivery. My main achievement was learning
+to work as a team.
+Projects
+Personalized Product Analysis with AI Python, Net, Huatei Cloud
+Oct 2023
+Our application is designed to help users make informed and healthy choices when purchasing products. By uploading a
+photo of the product's ingredients, the user can get a detailed analysis of how suitable and beneficial the product is for
+them. The application scans the content of the product with artificial intelligence systems and identifies substances that
+may cause allergies or adverse effects that the user has previously dentified. It provides the user with a summary of the
+product's content and the presence or absence of the substances they have identified. Thanks to this application, we
+came 3rd in BIk Academy and Huawei Coding Marathon
+Multi View Breast Cancer Classification App - Python, PyQs, Deep learming
+Apr 2022
+Within the scope of Teknofest artificial intelligence in health competition, the team I captained by developing a
+multi-model deep learning network for the diagnosis of breast cancers succeeded in becoming a finalist.
+Optical Character Recognition with Streamlit I Python, Streamlit, Huggingface
+Jan 2024
+OCR (Optical Character Recognition) technology has transformed how we interact with textual content in the digital
+realm. By converting images, scanned documents, and other media into editable and searchable text, OCR enables us to
+extract valuable information from diverse sources.
+
+--- Page 2 (DocTR) ---
+Technical Skills
+Languages: Python, C/ C++, Matlab, SQL, RobotStudio, Ros, PLC
+Developer Tools: Tensorflow, Pytorch, Google Cloud Platform, Huawei Cloud
+Technologies/Frameworks: Linux, GitHub, Selenium, Docker
+Certificates
+Teknofest Finalist Certificate:13 Foundation
+BTK Academy And Huawei Coding Marathon Certificate of Competitio Winning (3rd)
+EITCA Artificial Intelligence Academy 12 Certificates European Union
+TensorFlow: Advanced Techniques Specialization Coursera
+Google Cloud Expertise Google
+AI Expert Training Program 6 months Republic Of Tirkiye Ministry of Industry and Technology
+Introduction to Machine Learning in Production Coursera
+Hands-on ROS Training with Python :Udemy
+Image processing with deep learning :Udemy
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 9.17273736000061}
+OCR Result:
+--- Page 1 (DocTR) ---
+ALPEREN ÇELIK
+- +90 5453851876 peradkisfotgmalcom im.com/m/Aperenell-791aits, 0 llh-com/Alperencls,
+Education
+Afyon Kocatepe University
+Sep. 2018 - Jan 2024
+Bachelor of Mechatronic Engineering
+3.1 gpa
+Relevant Coursework
+Artificial Intelligence
+Software Methodology
+Database Management
+Internet Technology
+Computer Vision
+Algorithms Analysis
+Data Structures
+Systems Programming
+Experience
+Novelty AI
+Sep 2023 - Present
+AI/ML Engineer
+Gebze, Turkiye
+I developed dentification system software for a bank using deep learning techniques. This system aimed to increase
+security by optimizing customer authentication processes and was successfully implemented.
+At a defense industry company, I managed the installation of industrial robots (ABB) on the production line. In this
+project, I provided complex automation solutions to integrate robotic: systems and increase operational efficiency-
+For one of the leading telecom companies in Turkey, I developed software that enables live broadcasting and OTT
+automation with Suitest software using Python and image processing techniques.
+For a beverage company, I was part of the team that developed an artificial intelligence application that checks the
+recognition and accuracy of product labels on the production line.
+University of Malta
+Jul 2023 - Aug 2023 (3 mos)
+AI Researcher
+Maita
+* I worked on a sem-autonomous drone that tries to detect. waste pet bottles on beaches with artificial intelligence. I used
+Python and C++ languages in this project
+TC Diyanet isleri Bagkanhg
+May 2019 - Jul 2022 (3 yrs 3 mos)
+Civil Servant
+Ayonkarhisar, Turkiye
+While studying Mechatronics Engineering at the university, I worked as a civil servant and eammed a living. During these
+three years, the most essential value that this job added to me was to develop myself discipline and determination in
+order to keep the tough school and work life: in balance.
+DHMI Erzurum Airport
+Jul 2022 - Aug 2022 (2 mos)
+Mechatronics Engineer Intern
+Brzurum, Turkiye
+* I worked as an intern in areas such as sensors, electronic cards, x-ray devices in terminal electronics. My most significant
+gain from this internship was learning corporate work discipline and internal relationship techniques.
+Ecodation
+Jun 2021 - Jul 2021 (2 mos)
+Python Developer Intern
+Istanbul, Turkiye
+I made projects such as navigation and customer tracking system for cargo delivery. My main achievement was learning
+to work as a team.
+Projects
+Personalized Product Analysis with AI Python, Net, Huatei Cloud
+Oct 2023
+Our application is designed to help users make informed and healthy choices when purchasing products. By uploading a
+photo of the product's ingredients, the user can get a detailed analysis of how suitable and beneficial the product is for
+them. The application scans the content of the product with artificial intelligence systems and identifies substances that
+may cause allergies or adverse effects that the user has previously dentified. It provides the user with a summary of the
+product's content and the presence or absence of the substances they have identified. Thanks to this application, we
+came 3rd in BIk Academy and Huawei Coding Marathon
+Multi View Breast Cancer Classification App - Python, PyQs, Deep learming
+Apr 2022
+Within the scope of Teknofest artificial intelligence in health competition, the team I captained by developing a
+multi-model deep learning network for the diagnosis of breast cancers succeeded in becoming a finalist.
+Optical Character Recognition with Streamlit I Python, Streamlit, Huggingface
+Jan 2024
+OCR (Optical Character Recognition) technology has transformed how we interact with textual content in the digital
+realm. By converting images, scanned documents, and other media into editable and searchable text, OCR enables us to
+extract valuable information from diverse sources.
+
+--- Page 2 (DocTR) ---
+Technical Skills
+Languages: Python, C/ C++, Matlab, SQL, RobotStudio, Ros, PLC
+Developer Tools: Tensorflow, Pytorch, Google Cloud Platform, Huawei Cloud
+Technologies/Frameworks: Linux, GitHub, Selenium, Docker
+Certificates
+Teknofest Finalist Certificate:13 Foundation
+BTK Academy And Huawei Coding Marathon Certificate of Competitio Winning (3rd)
+EITCA Artificial Intelligence Academy 12 Certificates European Union
+TensorFlow: Advanced Techniques Specialization Coursera
+Google Cloud Expertise Google
+AI Expert Training Program 6 months Republic Of Tirkiye Ministry of Industry and Technology
+Introduction to Machine Learning in Production Coursera
+Hands-on ROS Training with Python :Udemy
+Image processing with deep learning :Udemy
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 9.556658029556274}
+OCR Result:
+--- Page 1 (DocTR) ---
+ALPEREN ÇELIK
+- +90 5453851876 peradkisfotgmalcom im.com/m/Aperenell-791aits, 0 llh-com/Alperencls,
+Education
+Afyon Kocatepe University
+Sep. 2018 - Jan 2024
+Bachelor of Mechatronic Engineering
+3.1 gpa
+Relevant Coursework
+Artificial Intelligence
+Software Methodology
+Database Management
+Internet Technology
+Computer Vision
+Algorithms Analysis
+Data Structures
+Systems Programming
+Experience
+Novelty AI
+Sep 2023 - Present
+AI/ML Engineer
+Gebze, Turkiye
+I developed dentification system software for a bank using deep learning techniques. This system aimed to increase
+security by optimizing customer authentication processes and was successfully implemented.
+At a defense industry company, I managed the installation of industrial robots (ABB) on the production line. In this
+project, I provided complex automation solutions to integrate robotic: systems and increase operational efficiency-
+For one of the leading telecom companies in Turkey, I developed software that enables live broadcasting and OTT
+automation with Suitest software using Python and image processing techniques.
+For a beverage company, I was part of the team that developed an artificial intelligence application that checks the
+recognition and accuracy of product labels on the production line.
+University of Malta
+Jul 2023 - Aug 2023 (3 mos)
+AI Researcher
+Maita
+* I worked on a sem-autonomous drone that tries to detect. waste pet bottles on beaches with artificial intelligence. I used
+Python and C++ languages in this project
+TC Diyanet isleri Bagkanhg
+May 2019 - Jul 2022 (3 yrs 3 mos)
+Civil Servant
+Ayonkarhisar, Turkiye
+While studying Mechatronics Engineering at the university, I worked as a civil servant and eammed a living. During these
+three years, the most essential value that this job added to me was to develop myself discipline and determination in
+order to keep the tough school and work life: in balance.
+DHMI Erzurum Airport
+Jul 2022 - Aug 2022 (2 mos)
+Mechatronics Engineer Intern
+Brzurum, Turkiye
+* I worked as an intern in areas such as sensors, electronic cards, x-ray devices in terminal electronics. My most significant
+gain from this internship was learning corporate work discipline and internal relationship techniques.
+Ecodation
+Jun 2021 - Jul 2021 (2 mos)
+Python Developer Intern
+Istanbul, Turkiye
+I made projects such as navigation and customer tracking system for cargo delivery. My main achievement was learning
+to work as a team.
+Projects
+Personalized Product Analysis with AI Python, Net, Huatei Cloud
+Oct 2023
+Our application is designed to help users make informed and healthy choices when purchasing products. By uploading a
+photo of the product's ingredients, the user can get a detailed analysis of how suitable and beneficial the product is for
+them. The application scans the content of the product with artificial intelligence systems and identifies substances that
+may cause allergies or adverse effects that the user has previously dentified. It provides the user with a summary of the
+product's content and the presence or absence of the substances they have identified. Thanks to this application, we
+came 3rd in BIk Academy and Huawei Coding Marathon
+Multi View Breast Cancer Classification App - Python, PyQs, Deep learming
+Apr 2022
+Within the scope of Teknofest artificial intelligence in health competition, the team I captained by developing a
+multi-model deep learning network for the diagnosis of breast cancers succeeded in becoming a finalist.
+Optical Character Recognition with Streamlit I Python, Streamlit, Huggingface
+Jan 2024
+OCR (Optical Character Recognition) technology has transformed how we interact with textual content in the digital
+realm. By converting images, scanned documents, and other media into editable and searchable text, OCR enables us to
+extract valuable information from diverse sources.
+
+--- Page 2 (DocTR) ---
+Technical Skills
+Languages: Python, C/ C++, Matlab, SQL, RobotStudio, Ros, PLC
+Developer Tools: Tensorflow, Pytorch, Google Cloud Platform, Huawei Cloud
+Technologies/Frameworks: Linux, GitHub, Selenium, Docker
+Certificates
+Teknofest Finalist Certificate:13 Foundation
+BTK Academy And Huawei Coding Marathon Certificate of Competitio Winning (3rd)
+EITCA Artificial Intelligence Academy 12 Certificates European Union
+TensorFlow: Advanced Techniques Specialization Coursera
+Google Cloud Expertise Google
+AI Expert Training Program 6 months Republic Of Tirkiye Ministry of Industry and Technology
+Introduction to Machine Learning in Production Coursera
+Hands-on ROS Training with Python :Udemy
+Image processing with deep learning :Udemy
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR", "Tesseract", "DocTR"], "Language": "English", "Device": "CPU", "Process Time": 31.933929920196533}
+OCR Result:
+--- Page 1 (DocTR) ---
+ALPEREN ÇELIK
+- +90 5453851876 peradkisfotgmalcom im.com/m/Aperenell-791aits, 0 llh-com/Alperencls,
+Education
+Afyon Kocatepe University
+Sep. 2018 - Jan 2024
+Bachelor of Mechatronic Engineering
+3.1 gpa
+Relevant Coursework
+Artificial Intelligence
+Software Methodology
+Database Management
+Internet Technology
+Computer Vision
+Algorithms Analysis
+Data Structures
+Systems Programming
+Experience
+Novelty AI
+Sep 2023 - Present
+AI/ML Engineer
+Gebze, Turkiye
+I developed dentification system software for a bank using deep learning techniques. This system aimed to increase
+security by optimizing customer authentication processes and was successfully implemented.
+At a defense industry company, I managed the installation of industrial robots (ABB) on the production line. In this
+project, I provided complex automation solutions to integrate robotic: systems and increase operational efficiency-
+For one of the leading telecom companies in Turkey, I developed software that enables live broadcasting and OTT
+automation with Suitest software using Python and image processing techniques.
+For a beverage company, I was part of the team that developed an artificial intelligence application that checks the
+recognition and accuracy of product labels on the production line.
+University of Malta
+Jul 2023 - Aug 2023 (3 mos)
+AI Researcher
+Maita
+* I worked on a sem-autonomous drone that tries to detect. waste pet bottles on beaches with artificial intelligence. I used
+Python and C++ languages in this project
+TC Diyanet isleri Bagkanhg
+May 2019 - Jul 2022 (3 yrs 3 mos)
+Civil Servant
+Ayonkarhisar, Turkiye
+While studying Mechatronics Engineering at the university, I worked as a civil servant and eammed a living. During these
+three years, the most essential value that this job added to me was to develop myself discipline and determination in
+order to keep the tough school and work life: in balance.
+DHMI Erzurum Airport
+Jul 2022 - Aug 2022 (2 mos)
+Mechatronics Engineer Intern
+Brzurum, Turkiye
+* I worked as an intern in areas such as sensors, electronic cards, x-ray devices in terminal electronics. My most significant
+gain from this internship was learning corporate work discipline and internal relationship techniques.
+Ecodation
+Jun 2021 - Jul 2021 (2 mos)
+Python Developer Intern
+Istanbul, Turkiye
+I made projects such as navigation and customer tracking system for cargo delivery. My main achievement was learning
+to work as a team.
+Projects
+Personalized Product Analysis with AI Python, Net, Huatei Cloud
+Oct 2023
+Our application is designed to help users make informed and healthy choices when purchasing products. By uploading a
+photo of the product's ingredients, the user can get a detailed analysis of how suitable and beneficial the product is for
+them. The application scans the content of the product with artificial intelligence systems and identifies substances that
+may cause allergies or adverse effects that the user has previously dentified. It provides the user with a summary of the
+product's content and the presence or absence of the substances they have identified. Thanks to this application, we
+came 3rd in BIk Academy and Huawei Coding Marathon
+Multi View Breast Cancer Classification App - Python, PyQs, Deep learming
+Apr 2022
+Within the scope of Teknofest artificial intelligence in health competition, the team I captained by developing a
+multi-model deep learning network for the diagnosis of breast cancers succeeded in becoming a finalist.
+Optical Character Recognition with Streamlit I Python, Streamlit, Huggingface
+Jan 2024
+OCR (Optical Character Recognition) technology has transformed how we interact with textual content in the digital
+realm. By converting images, scanned documents, and other media into editable and searchable text, OCR enables us to
+extract valuable information from diverse sources.
+
+--- Page 2 (DocTR) ---
+Technical Skills
+Languages: Python, C/ C++, Matlab, SQL, RobotStudio, Ros, PLC
+Developer Tools: Tensorflow, Pytorch, Google Cloud Platform, Huawei Cloud
+Technologies/Frameworks: Linux, GitHub, Selenium, Docker
+Certificates
+Teknofest Finalist Certificate:13 Foundation
+BTK Academy And Huawei Coding Marathon Certificate of Competitio Winning (3rd)
+EITCA Artificial Intelligence Academy 12 Certificates European Union
+TensorFlow: Advanced Techniques Specialization Coursera
+Google Cloud Expertise Google
+AI Expert Training Program 6 months Republic Of Tirkiye Ministry of Industry and Technology
+Introduction to Machine Learning in Production Coursera
+Hands-on ROS Training with Python :Udemy
+Image processing with deep learning :Udemy
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR", "DocTR", "Tesseract", "PaddleOCR"], "Language": "English", "Device": "CPU", "Process Time": 22.685651779174805}
+OCR Result:
+--- Page 1 (DocTR) ---
+RUNNING. Stop Deploy :
+Settings
+Select Device
+OCR and LLM Application
+e CPU
+O GPU (CUDA)
+Upload File (PDF, Image)
+V Save Outputs
+Drag and drop filel here
+Browse files
+Limit 200MB pert file-PDF,PNG, JPG, JPEG
+Selectlanguage
+English
+Select OCR Models
+EasyOCR X DOcTR x
+Tesseract x PaddleOcR
+SelectLLMN Model
+llama3.1
+Enter command:
+Selecttaskt type:
+e Summarize
+Generate
+
+
+---------------------------------------------------------------------------

results/ocr_output_EasyOCR.txt

@@ -0,0 +1,405 @@
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 0.6307122707366943}
+OCR Result:
+--- Page 1 (EasyOCR) ---
+Genel olarak sağ elinlz ne kadar iyl çalıştı?
+Sağ parmaklannız ne kadar iyi hareket etti?
+Sağ bileğiniz ne kadar iyi hareket etti?
+Sağ elinizin kuvveti nasıldı?
+Sağ elinizde
+(his) nasıldı?
+duyu
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR", "EasyOCR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 1.7959399223327637}
+OCR Result:
+--- Page 1 (EasyOCR) ---
+Genel olarak sağ elinlz ne kadar iyl çalıştı?
+Sağ parmaklannız ne kadar iyi hareket etti?
+Sağ bileğiniz ne kadar iyi hareket etti?
+Sağ elinizin kuvveti nasıldı?
+Sağ elinizde
+(his) nasıldı?
+duyu
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["DocTR", "EasyOCR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 1.8714756965637207}
+OCR Result:
+--- Page 1 (EasyOCR) ---
+Genel olarak sağ elinlz ne kadar iyl çalıştı?
+Sağ parmaklannız ne kadar iyi hareket etti?
+Sağ bileğiniz ne kadar iyi hareket etti?
+Sağ elinizin kuvveti nasıldı?
+Sağ elinizde
+(his) nasıldı?
+duyu
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR", "Tesseract", "DocTR"], "Language": "English", "Device": "CPU", "Process Time": 31.933929920196533}
+OCR Result:
+--- Page 1 (EasyOCR) ---
+ALPEREN CELIK
++yu ,isssisiu
+AerlS7Gtuicut
+AukELLOlllIW/Aluteu-celk-/414S1G
+Kthu CUt
+Aluerenclk /
+Educanou
+Afyon Kocatepe
+Universitu
+2U18
+Jan '2U244
+Becd-
+Wce
+EngmetinS
+ueau
+Cuursewors
+Lruncl
+Intelligence
+Sulttitt
+Methextlolugy
+DAlLc
+MatagCMLEut
+JcA
+Tecltolxy
+Cotupute
+Vsiol
+Algurithts
+AMNI:
+Dal Struceu
+XSLett
+PozHAing
+Experience
+Novelty
+Sep 2028
+Preemt
+AI{ML Engiutt'
+Gluzl _
+TuT(Jut
+veuc
+Ict ILLLcACLUM fsem   waEe
+USILE deek
+Jeautniug
+teciutles
+s * SLIIL MITL
+cELS
+JCurInI=
+#pLIMAZI= custumer authiettcatioln
+UEuC3s; Au Mi suclsslul;
+JHUeteteu
+l
+MT
+u
+Jalaged tle
+MsI;cM
+Faol
+(ABB
+tle productict Iitje _
+ptueci
+Doudrdouulex #uouabclnauutiuts
+E_HIC rulol I
+ALC MCEURC UPXCLUM]
+eflicietcy-
+Lalue
+CUELAAELC
+uk
+eveloue
+(at QHable=
+L; UEuiLSML A U
+MLUHAOU
+Suiest
+~utwit USlt= Puthutt
+IEMc DolFML [Lclelacc
+0Jikl
+UELAALY_
+develope
+ncllIM aa
+applicatin that checks
+MuuM
+Muaa
+mukluet lalels on the
+prouluction Iine .
+Universitv
+Malta
+Jul 2u23
+AuS 2023 (3 mos)
+Rec
+Multu
+WuLe
+FM umR Ca
+K[
+pet hottl:
+a
+Jck #ufc
+inte Illgetee .
+FBLI
+Fatl
+JitgMiyes
+prujec
+TC Diyanet isleri Daskanlgi
+Ni
+ZUlJ
+Jul 2022 (3 yra
+MJOSI
+Servuf (
+Auun UtJisu _
+TuT(Jut
+Wle STI]VL _eclulez-
+Eueeg
+MEOI;
+Wuke
+C ;CY[
+MVILE.
+113t
+Iiee tS_
+value that tils juls akdledl
+derJur [TAC dSCAEML"
+tleter uat IclL
+ut)
+Illt (ulleli *ol
+wutk Iit
+Dalce
+DHMI
+Erzunun
+Airport
+Jul 2U22
+2022
+MORI
+Auchutiue:
+Exeeet
+Iuter
+Erzu 4J{A
+TuT(Jut
+uckl
+ueru
+irLi
+ulSUs
+electrule LE
+X-t devices
+teHinal electronic My
+SeuicMHIC
+Ll IFUl
+[LS LcAMP %iatMl ColAlC FUk 4EWLC AEla
+IELTe ICAtlMaIL [CCMLUYR
+codarion
+2021
+Jul "Z
+Pedlu
+De
+K
+Ivux
+Tutiye
+cl
+MuCd
+TTLAl HL CU:[eE
+TACL >F1CM
+cargo delivery: My Wain achlereJleut #als
+Iemttg
+u
+M
+Frojects
+Peraqualked
+Pruduc
+Analysis wich AI
+Puk
+Net.
+Hnct Clatu
+203
+#pplication
+ataiuleu
+Qel USeI:
+Wfurtual
+M lt
+M
+Ge
+MECILLAJIY Decis
+By uplonaling #
+pelu
+tle potluct"
+iuguexlicuts,
+Uset GL 2el
+daraily #uales
+how" ?uitable aud Lxuefielal te petluct
+m
+The Applcaclon
+tlie [uuluet with Atificial
+intelllgenc:
+Mdemtln ? aulritce=
+C
+alletzie?
+e em
+Huuea
+Hleucife
+Doudas
+M MaT
+Rui
+product 
+T ' UFaa
+WR
+Liamuan
+Tlaul;
+thls applcaclon
+BTk Aeuety"
+Huaw ej Cucllug Marathon
+Mulce
+Vew
+Breast Cancer
+Clasalficatiun
+Pgthon. PyQtss
+Deep Jetc
+Apr 2022
+FtiL Ti' :UM"
+Tektulest artlcl] mtelluence
+lelth coletitull_
+CLiIL
+GuLilLd UY
+develjing
+Juulti-uudel deep leatuiug uetwak
+tlie dmeISI
+brenst Caucets sucurdtL
+MeLLI"
+HLA]st
+Uutical
+Charaeter Recoguition
+Aa
+Stremlit
+Pvn
+Staxurtt.
+Hadwefuc
+202a
+OCR
+Opeleal Characte
+Recuutil| [CCMUluy  Mis Iriledoned Mou
+teruc WIII textual cutett
+tle digital
+Jealu By cutvelting itage>
+TmMtL
+LUCIILICM3
+iLCumer MeU To €taule
+srclile tex_
+OCK e"alile?
+extract Wilu:LH
+MLLUlion
+ht dnvere auuce
+Sep:
+TeettLA
+coyet"
+JA 
+Dug
+AS Wl
+4u
+4ug
+kh
+404u
+uu
+
+--- Page 2 (EasyOCR) ---
+Technical Skills
+Languages:
+Fytlon €} C++-
+Matlal SQL.
+RukuStudl
+FLC
+Developer
+Tuula:
+Taue #lue
+Pytorch Guogl Clouel Flatfottu
+Huawel Cludl
+Techuologies
+autneworke
+Linux . GitHub. Selen_
+W
+Certilicates
+Tekuuleer
+Fluillat
+Cercileate T; Fouulaticn
+BTK
+ChdT
+eae
+Codlug
+Martlou
+Cetfc
+Cmpetitic Winuing (cd)
+EITCA
+Artaclal
+Intelligeuce
+ACha
+Certncates
+:Euro[an
+Uulu
+TeugorFlor
+Laten
+Teclulane
+Specializatlon
+Cuueari
+Google
+Cloud
+Expertise  Guogle
+Expert Traluing Program
+Womne
+Republic OF Tikiye Minisuty
+'Indlustiy
+Tehology
+Mr u
+Nacke
+Learmg
+Productou
+Cuuer
+Hauds-On ROS Tralniug
+Eitk
+Python :Udletny
+Itage processlug
+wtk dcer
+learuiug : Udlemy
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR", "DocTR", "Tesseract", "PaddleOCR"], "Language": "English", "Device": "CPU", "Process Time": 22.685651779174805}
+OCR Result:
+--- Page 1 (EasyOCR) ---
+RUNNING 
+Stop
+Deploy
+Settings
+Select Device
+OCR and LLM Application
+CPU
+GPU (CUDA)
+Upload File (PDF; Image)
+Drag and drop file here
+Save Outputs
+Browse files
+Limit ZOOMB per file : PDF; PNG; JPG, JPEG
+Select Language
+English
+Select OCR Models
+EasyOCR
+DocTR
+Tesseract
+PaddleOCR
+Select LLM Model
+Ilama3.1
+Enter command:
+Select task type:
+Summarize
+Generate
+
+
+---------------------------------------------------------------------------

results/ocr_output_PaddleOCR.txt

@@ -0,0 +1,259 @@
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["PaddleOCR"], "Language": "English", "Device": "CPU", "Process Time": 13.193224430084229}
+OCR Result:
+--- Page 1 (PaddleOCR) ---
+ALPEREN
+CELIK
++90 5453851876
+ alperenclk18760gmail.cominlinkedin.com/in/alperen-celik-7919a5163/
+ github.com/Alperenclk/
+Education
+Afyon Kocatepe University
+Sep. 2018  Jan 2024
+Bachelor of Mechatronic Engineering
+3.1 gpa
+Relevant Coursework
+ Artificial Intelligence
+ Software Methodology
+ Database Management
+ Internet Technology
+ Computer Vision
+ Algorithms Analysis
+ Data Structures
+ Systems Programming
+Experience
+Novelty AI
+Sep 2023  Present
+AI/ML Engineer
+Gebze, Turkiye
+ I developed identification system software for a bank using deep learning techniques. This system aimed to increase
+security by optimizing customer authentication processes and was successfully implemented.
+At a defense industry company, I managed the installation of industrial robots (ABB) on the production line. In this
+project, I provided complex automation solutions to integrate robotic systems and increase operational efficiency.
+For one of the leading telecom companies in Turkey, I developed software that enables live broadcasting and OTT
+automation with Suitest software using Python and image processing techniques.
+ For a beverage company, I was part of the team that developed an artificial intelligence application that checks the
+recognition and accuracy of product labels on the production line.
+University of Malta
+Jul 2023  Aug 2023 (3 mos)
+AI Researcher
+Malt
+I worked on a semi-autonomous drone that tries to detect waste pet bottles on beaches with artificial intelligence. I used
+Python and C++ languages in this project
+TC Diyanet Isleri Baskanlig1
+May 2019  Jul 2022 (3 yrs 3 mos)
+Civil Servant
+Afyonkarhisar, Turkiye
+ While studying Mechatronics Engineering at the university, I worked as a civil servant and earned a living. During these
+three years, the most essential value that this job added to me was to develop myself discipline and determination in
+order to keep the tough school and work life in balance.
+DHMI Erzurum Airport
+Jul 2022  Aug 2022 (2 mos)
+Mechatronics Engineer Intern
+Erzurum, Turkiye
+ I worked as an intern in areas such as sensors, electronic cards, x-ray devices in terminal electronics. My most significant
+gain from this internship was learning corporate work discipline and internal relationship techniques.
+Ecodation
+Jun 2021  Jul 2021 (2 mos)
+Python Developer Intern
+Istanbul, Turkiye
+ I made projects such as navigation and customer tracking system for cargo delivery. My main achievement was learning
+to work as a team.
+Projects
+Personalized Product Analysis with AI | Python, .Net, Huawei Cloud
+Oct 2023
+ Our application is designed to help users make informed and healthy choices when purchasing products. By uploading a
+photo of the product's ingredients, the user can get a detailed analysis of how suitable and beneficial the product is for
+them. The application scans the content of the product with artificial intelligence systems and identifies substances that
+may cause allergies or adverse effects that the user has previously identified. It provides the user with a summary of the
+product's content and the presence or absence of the substances they have identified. Thanks to this application, we
+came 3rd in BTk Academy and Huawei Coding Marathon
+Multi View Breast Cancer Classification App | Python, PyQt5, Deep learning
+Apr 2022
+ Within the scope of Teknofest artificial intelligence in health competition, the team I captained by developing a
+multi-model deep learning network for the diagnosis of breast cancers succeeded in becoming a finalist.
+Optical Character Recognition with Streamlit | Python, Streamlit, Huggingface
+Jan 2024
+ OCR (Optical Character Recognition) technology has transformed how we interact with textual content in the digital
+realm. By converting images, scanned documents, and other media into editable and searchable text, OCR enables us to
+extract valuable information from diverse sources.
+
+--- Page 2 (PaddleOCR) ---
+Technical Skills
+Languages: Python, C/ C++, Matlab, SQL, RobotStudio, Ros, PLC
+Developer Tools: Tensorflow, Pytorch, Google Cloud Platform, Huawei Cloud
+Technologies/Frameworks: Linux, GitHub, Selenium, Docker
+Certificates
+Teknofest Finalist Certificate:T3 Foundation
+BTK Academy And Huawei Coding Marathon :Certificate of Competitio Winning (3rd)
+EITCA Artificial Intelligence Academy 12 Certificates :European Union
+TensorFlow: Advanced Techniques Specialization :Coursera
+Google Cloud Expertise :Google
+AI Expert Training Program 6 months :Republic Of Tirkiye Ministry of Industry and Technology
+Introduction to Machine Learning in Production :Coursera
+Hands-on ROS Training with Python :Udemy
+Image processing with deep learning :Udemy
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["PaddleOCR"], "Language": "Türkçe", "Device": "CPU", "Process Time": 6.291873216629028}
+OCR Result:
+--- Page 1 (PaddleOCR) ---
+ALPEREN QELIK
+ +90 5453851876
+ eng.alperengmail.com
+in linkedin.com/in/alperen-celik-7919a5163/
+ github.com/Alperenclk/
+Education
+Afyon Kocatepe University
+Sep. 2018  Jan 2024
+Bachelor of Mechatronic Engincering
+3.1 gpa
+Relevant Coursework
+ Artificial Intelligence
+ Robotics
+ Machine Learning
+ Computer Vision
+ Cloud Systems
+ Deep Learning
+ NLP
+ Database Management
+Experience
+Novelty AI
+Sep 2023  Present
+AI/ML Engincer
+GebzcTurkiye
+ Using advanced Computer Vision and deep learning techniques, I have developed a system that develops billboards that
+provide personalized advertising by analyzing the characteristics of customers who come to stores in shopping malls. This
+system works 15% more successfully than the foreign product previously used.
+ Developed C# software using profile sensors from Sick and Venglor to control parts production for a Japan-based factory
+With this software, we used advanced image processing techniques to process and analyze point cloud data to accurately
+monitor the production process and detect defective parts, increasing the success rate of production to 95%.°
+ I developed identification system software for a bank using deep learning techniques. This system aimed to increase
+security by optimizing customer authentication processes and was successfully implemented.
+_At a defense industry company, I led the installation of ABB industrial robots on a production line. Leveraging advanced
+Computer Vision techniques, I developed a system that commands robots to perform tasks traditionally performed by
+human operators. This project used Artificial Intelligence and robotics to deliver complex automation solutions, greatly
+improving operational efficiency.
+ For one of the leading telecom companies in Turkey, I developed software that enables live broadcasting and OTT
+automation with Suitest software using Python and image processing techniques.
+ For a beverage company, I was part of the team that developed an artificial intelligence application that checks the
+recognition and accuracy of product labels on the production line.
+University of Malta
+Jul 2023  Aug 2023 (3 mos)
+Al Rescarcher
+Malta
+ I worked with a team developing a semi-autonomous drone to detect waste plastic bottles on beaches using artificial
+intelligence. I implemented computer vision algorithms such as Detectron2 in Python for object detection and used C++
+for real-time integration with the drone's control systems, enabling efficient environmental scanning and cleaning.
+DHMI Erzurum Airport
+Jul 2022  Aug 2022 (2 mos)
+Mechatronics Engincer Intern
+Erzurum, Tirkiye
+ I interned in areas such as sensors, electronic cards, and x-ray devices within terminal electronics. The most significant
+gain from this internship was learning corporate work discipline and internal relationship techniques..
+Ecodation
+Jun 2021  Jul 2021 (2 mos)
+Python Developer Intern
+Istanul,Trkiy
+ I worked on projects like a navigation and customer tracking system for cargo delivery. These projects involved
+developing and integrating Python APIs and Flask to streamline data communication and using web scraping techniques
+to gather and analyze relevant information from various sources. My main achievement from these projects was learning
+to work effectively as a team, coordinating with colleagues to tackle complex challenges and deliver cohesive solutions.
+
+--- Page 2 (PaddleOCR) ---
+Projects
+AI-Driven Cryptocurrency Trading Bot Python, LLM, LangChain, AI Agents
+ Developed a cryptocurrency trading bot utilizing Large Language Models (LLMs) and LangChain to enhance trading
+decisions. The bot integrates with the Binance API, executing trades based on real-time market data and technical
+analysis.
+ The LangChain framework and AI agents are used to analyze market trends by combining historical data with current
+information. This setup enables the bot to perform detailed technical analysis and generate actionable insights for
+trading.
+ AI agents process news and social media to assess market sentiment. The bot adjusts its trading strategies based on
+sentiment and technical indicators, aiming to maximize profitability and minimize risk.
+ Python was used for developing the core functionalities and integrating the LangChain system, ensuring the bot's
+efficiency and effectiveness in the volatile cryptocurrency market. The bot also adapts and improves its strategies over
+time through continuous learning.
+Personalized Product Analysis with AI  Python, .Net, Huauei Cloud
+ Our application is designed to help users make informed and healthy choices when purchasing products. By uploading a
+photo of the product's ingredients, the user can get a detailed analysis of how suitable and beneficial the product is for
+them. The application scans the content of the product with artificial intelligence systems and identifies substances that
+may cause allergies or adverse effects that the user has previously identified. It provides the user with a summary of the
+product's content and the presence or absence of the substances they have identified. Thanks to this application, we
+came 3rd in BTk Academy and Huawei Coding Marathon
+Multi View Breast Cancer Classification App  Python, PyQt5, Deep Iearning
+ Within the scope of Teknofest artificial intelligence in health competition, the team I captained by developing a
+multi-model deep learning network for the diagnosis of breast cancers succeeded in becoming a finalist.
+Optical Character Recognition with Streamlit  Python, Streamlit, Huggingface
+_OCR (Optical Character Recognition) technology has transformed how we interact with textual content in the digital
+realm. By converting images, scanned documents, and other media into editable and searchable text, OCR enables us to
+extract valuable information from diverse sources.
+Technical Skills
+Languages: Python, C/ C++, Matlab, SQL, RobotStudio, Ros, PLC
+Developer Tools: Tensorfiow, Pytorch, Google Cloud Platform, Huawei Cloud
+Technologies/Frameworks: Linux, GitHub, HuggingFace, Kaggle, Selenium, Docker
+Certificates
+Teknofest Finalist Certificate: T3 Foundation
+BTK Academy And Huawei Coding Marathon : Certificate of Competitio Winning (3rd)
+EITCA Artificial Intelligence Academy 12 Certificates : European Union
+TensorFlow: Advanced Techniques Specialization : Coursera
+Google Cloud Expertise : Google
+AI Expert Training Program 6 months_ : Republic Of Tirkiye Ministry of Industry and Technology
+Oxford University English B2 Certification : ClubClass Language School MALTA
+Introduction to Machine Learning in Production : Coursera
+Hands-on ROS Training with Python : Udemy
+Image processing with Deep learning : Udemy
+Honors and Open Source Pojects
+-Teknofest Healthcare Competition Finalist
+-BTK Academy and Huawei Coder Marathon Third Place
+-Kaggle Master https://www.kaggle.com/alperenclk
+Some Articles
+https://medium.com/@alperenclk/exploring-optical-character-recognition-ocr-with-streamlit-and-doctr-
+00e95ae36e4e
+https://medium.com/@alperenclk/automating-telegram-game-bot-clicker-with-python-step-by-step-guide-
+1b9206188d06
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR", "DocTR", "Tesseract", "PaddleOCR"], "Language": "English", "Device": "CPU", "Process Time": 22.685651779174805}
+OCR Result:
+--- Page 1 (PaddleOCR) ---
+ RUNNING...
+Stop
+ Deploy
+ Settings
+Select Device
+OCR and LLM Application
+O CPU
+Upload File (PDF, Image)
+O GPU (CUDA)
+Save Outputs
+4
+Drag and drop file here.
+Browse files
+imit 200MB per file - PDF, PNG, JPG, JPEG
+Select Language
+ English
+Select OCR Models
+EasyOCR x
+DocTR x
+Tesseract 
+ PaddleOCR x
+Select LLM Mode!
+ llama3.1
+Enter command:
+Select task type:
+O Summarize
+O Generate
+
+
+---------------------------------------------------------------------------

results/ocr_output_Tesseract.txt

@@ -0,0 +1,146 @@
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR", "Tesseract", "DocTR"], "Language": "English", "Device": "CPU", "Process Time": 31.933929920196533}
+OCR Result:
+--- Page 1 (Tesseract) ---
+ALPEREN CELIK
+
+7-400 5453861875 WB alperoneSTONguil.com inked com/in/alperen-ell-TO95163/€ github, com/ Alperenelk/
+
+Education
+‘Afyon Kocatepe University Sep. 2018 — Jan 2024
+Bachelor of Mechatronic Engineering 3 gpa
+
+Relevant Coursework
+
++ Artifial nteligence + Software Methodology + Database Management» Internet Technology
+2 Cimputer Vision 1 Algartinas Analyste 1 bata Structuses 1 Syatens Progeamtning
+
+Experience
+
+Novelty AI Sep 2028 — Present
+
+AU/ML Engineer Gebse, Tinaye
+
++ Tdeveloped ientication system software fora bank using deep learning techniques. This system aimed to inrease
+sccuit by optimizing eustomer authentication proceses and was sucessfully implemented
+
++ Ata defense industry company, I managed the installation of industrial robots (ABB) om the production tine, In this
+projet, Tprovided complex automation solutions to integrate robotic systems and increase operational eBclene.
+
++ For one of the leading telecom companies in Turkey, I developed software that enable ive broadcasting and OTT
+automation with Sutest software using Python and image procesing techniques
+
++ For a beverage company, Iwas port of the team that developed an artical inteligence application that checks the
+recognition and accuracy of product labels on the production ine
+
+University of Malta Jul 2023 ~ Aug 2028 (3 mos)
+
+AI Researcher Matte
+
++ T worked on @ semi-autonomous drone that Wis to detect waste pet bottles on beads with artical intelligence Y used
+Python and C+ languages inthis project
+
+May 2019 — Jul 2022 (8 yes mos)
+ini Servant Afyonkarhisar, Mirksye
++ While studying Mechatvoles Engineering atthe university, I worked as a ull servant and earned a living. During these
+
+‘htee years, the most essential value that this job added to me was to develop tnyelfdlacplne and detrtnation in
+fonder to keep the tough schol and work fe tn balance
+
+DHMI Erzurum Airport Jul 2022 ~ Aug 2022 (2 mos)
+
+Mechatwonses Engineer Intern Brsurumn, Tirkiye
+
++ T worked an inter in azeas such as sensors electronic cards, s-zay deviees in terminal eeetronies. My most significant
+sain from this interasip was leaening corporate work dicpline and internal yelationship techniques
+
+Ecodation Jun 2021 ~ Jul 2021 (2 mos)
+Python Developer Intern Ibtantad, Taye
+
++ Tmade projects uch as navigation and customer tracking system fr cargo delivery. My main achievement was learning
+to wotk asa teat,
+
+Projects
+
+Personalized Product Analysis with AL| Python, Net, Huawei Clout (Oct 2028
++ Our appleation x designed to help users make informed and healthy choloss when purchasing products, By uploading a
+photo ofthe product's ingredients the user can get detaled analysis of how stable and beneficial the product is for
+‘hem, The application scans the content of the produet with aetiial intelligence systems and identifies substances that
+say cause allergis or adverse effects that the user has previously identified. Te provides the user with a summary’ of the
+‘Product's coutent and the presence or absence ofthe substances they have identified, Thanks to this application, we
+fame Sed in BT Academy and Huawel Coding Marathon
+
+‘Multi View Breast Cancer Classification App | Python, PyQU, Deep loaning Ape 2022
++ Within the scope of Teknofest artical intligence in heath competition, the team T captained ly developing a
+sultimodel dep learning network forthe diagnosis of breast cancers succeded in becoming Bnalst,
+
+Optical Character Recognition with Streamlt | Python, Steamtt, Hagningface Jan 2024
++ OCR (Optical Character Recogution) technology has transformed how we interact with textual content ia the digital
+realm By converting images, seanned documents, and other media into editable and searchable txt, OCR enables Us to
+‘extract valuable infrination from diverse sourecs.
+
+
+--- Page 2 (Tesseract) ---
+‘Technical Skills
+
+Languages: Python, C/ C++, Matlab, SQL, RobotStudlo, Ros, PLE
+Developer Tools: Tensoiow. Pytore Google Cloud Platfnn, Huawel Cloud
+‘Technologies Frameworks: Linus, GitHub, Selenium, Docker
+
+Certificates
+
+‘Telofest Finalist CertfieatecT3 Foundation
+BTK Academy And Huawel Coding Marathon -Certiieate of Compatitio Wining (Sed)
+EITCA Artificial Intelligence Academy 12 Certificates European Union
+
+‘TensorFlow: Advanced Techniques Speciallaation :Coursera
+
+Google Cloud Expertise :G
+
+AT Expert Training Progr dhs. -Republic OF Tiskiye Ministry of Industry and Technology
+Introduction to Machine Learning in Production :Coutsers
+
+Hands-on ROS Training with Python idem
+
+Image processing with deep learning -Udeny
+
+
+
+
+---------------------------------------------------------------------------
+
+---------------------------------------------------------------------------
+Attributes of Output:
+{"Model Names": ["EasyOCR", "DocTR", "Tesseract", "PaddleOCR"], "Language": "English", "Device": "CPU", "Process Time": 22.685651779174805}
+OCR Result:
+--- Page 1 (Tesseract) ---
+< RUNNING... Stop Deploy
+
+Settings
+
+covert OCR and LLM Application
+
+@ cpu
+GPU (CUDA)
+
+@ Save Outputs
+
+Select Language
+English v
+Select OCR Models
+
+Eee Ene
+(Tesseract x) [Paduleoce x) © ~
+
+‘Select LLM Model
+Enter command:
+Select task type:
+
+® Summarize
+Generate
+
+
+
+---------------------------------------------------------------------------

sample_files/alperen_celik_14_08.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:700730cfc0595417324f787e2ad30b96763e0b551c95717e875d8e2191b6ebe9
+size 131864

streamlit_app.py

@@ -0,0 +1,201 @@
+"""Alias entrypoint for Streamlit on Hugging Face Spaces.
+This is a copy of app.py to match Spaces' default file naming.
+"""
+
+import streamlit as st
+from PIL import Image
+import fitz  # PyMuPDF
+import numpy as np
+import tempfile
+import os
+import time
+import io
+import json
+import torch
+import cv2
+
+# Import OCR engines
+import ocr_engines
+
+# Try importing LLM processor if LLM features are to be used
+llm_available = False
+try:
+    import llm_processor
+
+    llm_available = True
+except ImportError:
+    pass  # LLM features will be disabled
+
+# Create results folder if it doesn't exist
+if not os.path.exists("results"):
+    os.makedirs("results")
+
+# Streamlit application
+st.title("OCRInsight")
+
+# Sidebar
+st.sidebar.header("Settings")
+
+
+# Function to save text to file
+def save_text_to_file(attributes_of_output, all_ocr_text, filename):
+    with open(filename, "a", encoding="utf-8") as f:
+        f.write("\n" + "-" * 75 + "\n")
+        f.write("Attributes of Output:\n")
+        f.write(attributes_of_output)
+        f.write("\nOCR Result:\n")
+        f.write(all_ocr_text)
+        f.write("\n" + "-" * 75 + "\n")
+    st.success(f"{filename} saved successfully!")
+
+
+# Device selection
+device = st.sidebar.radio("Select Device", ["CPU", "GPU (CUDA)"])
+save_output = st.sidebar.checkbox("Save Outputs")
+
+# Language selection
+language = st.sidebar.selectbox(
+    "Select Language", ["Türkçe", "English", "Français", "Deutsch", "Español"]
+)
+
+# Map selected language to language codes
+language_codes = {
+    "Türkçe": "tr",
+    "English": "en",
+    "Français": "fr",
+    "Deutsch": "de",
+    "Español": "es",
+}
+
+# OCR model selection
+ocr_models = st.sidebar.multiselect(
+    "Select OCR Models",
+    ["EasyOCR", "DocTR", "Tesseract", "PaddleOCR"],
+    ["EasyOCR"],  # default selection
+)
+
+# LLM model selection
+llm_model = st.sidebar.selectbox(
+    "Select LLM Model", ["Only OCR Mode", "llama3.1", "llama3", "gemma2"]
+)
+
+# Conditional UI elements based on LLM model selection
+if llm_model != "Only OCR Mode" and llm_available:
+    user_command = st.sidebar.text_input("Enter command:", "")
+
+    task_type = st.sidebar.radio("Select task type:", ["Summarize", "Generate"])
+elif llm_model != "Only OCR Mode" and not llm_available:
+    st.sidebar.warning(
+        "LLM features are not available. Please install 'ollama' to enable LLM processing."
+    )
+    llm_model = "Only OCR Mode"
+
+# Check GPU availability
+if device == "GPU (CUDA)" and not torch.cuda.is_available():
+    st.sidebar.warning("GPU (CUDA) not available. Switching to CPU.")
+    device = "CPU"
+
+# Initialize OCR models
+ocr_readers = ocr_engines.initialize_ocr_models(
+    ocr_models, language_codes[language], device
+)
+
+# File upload
+uploaded_file = st.file_uploader(
+    "Upload File (PDF, Image)", type=["pdf", "png", "jpg", "jpeg"]
+)
+
+# Create results folder if it doesn't exist
+if not os.path.exists("results"):
+    os.makedirs("results")
+
+if uploaded_file is not None:
+    start_time = time.time()
+
+    if uploaded_file.type == "application/pdf":
+        pdf_document = fitz.open(stream=uploaded_file.read(), filetype="pdf")
+        images = []
+        for page_num in range(len(pdf_document)):
+            page = pdf_document.load_page(page_num)
+            pix = page.get_pixmap()
+            img_data = pix.tobytes("png")
+            img = Image.open(io.BytesIO(img_data))
+            images.append(img)
+        total_pages = len(pdf_document)
+        pdf_document.close()
+    else:
+        images = [Image.open(uploaded_file)]
+        total_pages = 1
+
+    all_ocr_texts = {
+        model_name: "" for model_name in ocr_models
+    }  # To store OCR text for each model
+
+    for page_num, image in enumerate(images, start=1):
+        st.image(image, caption=f"Page {page_num}/{total_pages}", use_column_width=True)
+
+        # Perform OCR with each selected model
+        for model_name in ocr_models:
+            text = ocr_engines.perform_ocr(
+                model_name, ocr_readers, image, language_codes[language]
+            )
+            all_ocr_texts[
+                model_name
+            ] += f"--- Page {page_num} ({model_name}) ---\n{text}\n\n"
+
+            st.subheader(f"OCR Result ({model_name}) - Page {page_num}/{total_pages}:")
+            st.text(text)
+
+    end_time = time.time()
+    process_time = end_time - start_time
+
+    st.info(f"Processing time: {process_time:.2f} seconds")
+
+    # Save OCR outputs if selected
+    if save_output:
+        attributes_of_output = {
+            "Model Names": ocr_models,
+            "Language": language,
+            "Device": device,
+            "Process Time": process_time,
+        }
+        for model_name, ocr_text in all_ocr_texts.items():
+            filename = f"results//ocr_output_{model_name}.txt"
+            save_text_to_file(
+                json.dumps(attributes_of_output, ensure_ascii=False), ocr_text, filename
+            )
+
+    # LLM processing
+    if (
+        llm_model != "Only OCR Mode"
+        and llm_available
+        and st.sidebar.button("Start LLM Processing")
+    ):
+        st.subheader("LLM Processing Result:")
+
+        # Combine all OCR texts
+        combined_ocr_text = "\n".join(all_ocr_texts.values())
+
+        # Prepare the prompt based on the task type
+        if task_type == "Summarize":
+            prompt = f"Please summarize the following text. Command: {user_command}\n\nText: {combined_ocr_text}"
+        else:  # "Generate"
+            prompt = f"Please generate new text based on the following text. Command: {user_command}\n\nText: {combined_ocr_text}"
+
+        llm_output = llm_processor.process_with_llm(llm_model, prompt)
+
+        # Display the result
+        st.write(f"Processing completed using '{llm_model}' model.")
+        st.text_area("LLM Output:", value=llm_output, height=300)
+
+        # Save LLM output if selected
+        if save_output:
+            filename = "llm_output.txt"
+            save_text_to_file(llm_output, "", filename)
+
+elif llm_model != "Only OCR Mode" and not llm_available:
+    st.warning(
+        "LLM features are not available. Please install 'ollama' to enable LLM processing."
+    )
+
+st.sidebar.info(f"Selected device: {device}")