Upload app.py with huggingface_hub
Browse files
app.py
ADDED
|
@@ -0,0 +1,282 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import numpy as np
|
| 3 |
+
from PIL import Image, ImageDraw, ImageFont
|
| 4 |
+
import io
|
| 5 |
+
import os
|
| 6 |
+
from typing import Optional, Tuple, List
|
| 7 |
+
import re
|
| 8 |
+
|
| 9 |
+
# Mock OCR function (in real implementation, you'd use pytesseract or similar)
|
| 10 |
+
def detect_text_in_image(image: np.ndarray) -> List[str]:
|
| 11 |
+
"""Mock function to detect existing text in image"""
|
| 12 |
+
# In a real implementation, this would use OCR like pytesseract
|
| 13 |
+
# For demo purposes, we'll return some sample detected text
|
| 14 |
+
return ["Sample text detected", "Another text element"]
|
| 15 |
+
|
| 16 |
+
def add_text_to_image(
|
| 17 |
+
image: np.ndarray,
|
| 18 |
+
new_text: str,
|
| 19 |
+
text_position: str,
|
| 20 |
+
font_size: int,
|
| 21 |
+
text_color: str,
|
| 22 |
+
background_color: str,
|
| 23 |
+
detect_existing: bool,
|
| 24 |
+
accessibility_description: str
|
| 25 |
+
) -> Tuple[np.ndarray, str, str]:
|
| 26 |
+
"""
|
| 27 |
+
Add text overlay to image with accessibility features
|
| 28 |
+
"""
|
| 29 |
+
if image is None:
|
| 30 |
+
return None, "Please upload an image first", ""
|
| 31 |
+
|
| 32 |
+
# Convert numpy array to PIL Image
|
| 33 |
+
pil_image = Image.fromarray(image.astype('uint8'), 'RGB')
|
| 34 |
+
|
| 35 |
+
# Detect existing text if requested
|
| 36 |
+
existing_text_info = ""
|
| 37 |
+
if detect_existing:
|
| 38 |
+
try:
|
| 39 |
+
detected_texts = detect_text_in_image(image)
|
| 40 |
+
if detected_texts:
|
| 41 |
+
existing_text_info = "Detected existing text:\n" + "\n".join(f"• {text}" for text in detected_texts)
|
| 42 |
+
else:
|
| 43 |
+
existing_text_info = "No existing text detected in the image."
|
| 44 |
+
except Exception as e:
|
| 45 |
+
existing_text_info = f"Text detection failed: {str(e)}"
|
| 46 |
+
|
| 47 |
+
# Add new text overlay
|
| 48 |
+
draw = ImageDraw.Draw(pil_image)
|
| 49 |
+
|
| 50 |
+
# Create font (you might want to use a proper font file in production)
|
| 51 |
+
try:
|
| 52 |
+
font = ImageFont.truetype("arial.ttf", font_size)
|
| 53 |
+
except:
|
| 54 |
+
font = ImageFont.load_default()
|
| 55 |
+
|
| 56 |
+
# Get image dimensions
|
| 57 |
+
img_width, img_height = pil_image.size
|
| 58 |
+
|
| 59 |
+
# Calculate text position
|
| 60 |
+
text_bbox = draw.textbbox((0, 0), new_text, font=font)
|
| 61 |
+
text_width = text_bbox[2] - text_bbox[0]
|
| 62 |
+
text_height = text_bbox[3] - text_bbox[1]
|
| 63 |
+
|
| 64 |
+
position_map = {
|
| 65 |
+
"top-left": (10, 10),
|
| 66 |
+
"top-center": ((img_width - text_width) // 2, 10),
|
| 67 |
+
"top-right": (img_width - text_width - 10, 10),
|
| 68 |
+
"center-left": (10, (img_height - text_height) // 2),
|
| 69 |
+
"center": ((img_width - text_width) // 2, (img_height - text_height) // 2),
|
| 70 |
+
"center-right": (img_width - text_width - 10, (img_height - text_height) // 2),
|
| 71 |
+
"bottom-left": (10, img_height - text_height - 10),
|
| 72 |
+
"bottom-center": ((img_width - text_width) // 2, img_height - text_height - 10),
|
| 73 |
+
"bottom-right": (img_width - text_width - 10, img_height - text_height - 10)
|
| 74 |
+
}
|
| 75 |
+
|
| 76 |
+
x, y = position_map.get(text_position, position_map["bottom-center"])
|
| 77 |
+
|
| 78 |
+
# Add background rectangle for better readability (optional)
|
| 79 |
+
if background_color != "transparent":
|
| 80 |
+
# Convert hex color to RGB
|
| 81 |
+
if background_color.startswith('#'):
|
| 82 |
+
bg_rgb = tuple(int(background_color[i:i+2], 16) for i in (1, 3, 5))
|
| 83 |
+
else:
|
| 84 |
+
bg_rgb = (0, 0, 0) # default black
|
| 85 |
+
|
| 86 |
+
# Draw background rectangle
|
| 87 |
+
padding = 5
|
| 88 |
+
draw.rectangle([
|
| 89 |
+
x - padding, y - padding,
|
| 90 |
+
x + text_width + padding, y + text_height + padding
|
| 91 |
+
], fill=bg_rgb)
|
| 92 |
+
|
| 93 |
+
# Convert text color to RGB
|
| 94 |
+
if text_color.startswith('#'):
|
| 95 |
+
text_rgb = tuple(int(text_color[i:i+2], 16) for i in (1, 3, 5))
|
| 96 |
+
else:
|
| 97 |
+
text_rgb = (255, 255, 255) # default white
|
| 98 |
+
|
| 99 |
+
# Draw text
|
| 100 |
+
draw.text((x, y), new_text, fill=text_rgb, font=font)
|
| 101 |
+
|
| 102 |
+
# Convert back to numpy array
|
| 103 |
+
result_image = np.array(pil_image)
|
| 104 |
+
|
| 105 |
+
# Generate accessibility description
|
| 106 |
+
if not accessibility_description:
|
| 107 |
+
accessibility_description = f"Image with text overlay: '{new_text}' positioned at {text_position}"
|
| 108 |
+
|
| 109 |
+
# Combine all information for accessibility
|
| 110 |
+
full_accessibility_info = f"Accessibility Description: {accessibility_description}\n\n"
|
| 111 |
+
if existing_text_info:
|
| 112 |
+
full_accessibility_info += f"{existing_text_info}\n\n"
|
| 113 |
+
full_accessibility_info += f"Added text: '{new_text}' at {text_position} with font size {font_size}"
|
| 114 |
+
|
| 115 |
+
return result_image, full_accessibility_info, existing_text_info
|
| 116 |
+
|
| 117 |
+
# Create the Gradio interface
|
| 118 |
+
with gr.Blocks(title="Premium Image Text Overlay - Accessibility Edition") as demo:
|
| 119 |
+
gr.HTML("""
|
| 120 |
+
<div style="text-align: center; margin-bottom: 20px;">
|
| 121 |
+
<h1 style="color: #4A90E2; font-size: 2.5em;">🖼️ Premium Image Text Overlay</h1>
|
| 122 |
+
<p style="font-size: 1.2em; color: #666;">Enhanced accessibility for visually impaired users with text detection</p>
|
| 123 |
+
<p><a href="https://huggingface.co/spaces/akhaliq/anycoder" target="_blank">Built with anycoder</a></p>
|
| 124 |
+
</div>
|
| 125 |
+
""")
|
| 126 |
+
|
| 127 |
+
with gr.Tabs():
|
| 128 |
+
with gr.Tab("🖼️ Image Editor"):
|
| 129 |
+
with gr.Row():
|
| 130 |
+
with gr.Column():
|
| 131 |
+
input_image = gr.Image(
|
| 132 |
+
label="Upload Image",
|
| 133 |
+
type="numpy",
|
| 134 |
+
height=400,
|
| 135 |
+
interactive=True
|
| 136 |
+
)
|
| 137 |
+
|
| 138 |
+
with gr.Accordion("🔍 Text Detection Settings", open=True):
|
| 139 |
+
detect_existing = gr.Checkbox(
|
| 140 |
+
label="Detect existing text in image",
|
| 141 |
+
value=True,
|
| 142 |
+
info="Automatically detect and display any text already present in the uploaded image"
|
| 143 |
+
)
|
| 144 |
+
|
| 145 |
+
existing_text_display = gr.Textbox(
|
| 146 |
+
label="Existing Text Detection Results",
|
| 147 |
+
interactive=False,
|
| 148 |
+
max_lines=5,
|
| 149 |
+
placeholder="Existing text will appear here after detection..."
|
| 150 |
+
)
|
| 151 |
+
|
| 152 |
+
with gr.Column():
|
| 153 |
+
output_image = gr.Image(
|
| 154 |
+
label="Result Image",
|
| 155 |
+
type="numpy",
|
| 156 |
+
height=400,
|
| 157 |
+
interactive=False
|
| 158 |
+
)
|
| 159 |
+
|
| 160 |
+
accessibility_output = gr.Textbox(
|
| 161 |
+
label="♿ Accessibility Information",
|
| 162 |
+
interactive=False,
|
| 163 |
+
max_lines=8,
|
| 164 |
+
placeholder="Comprehensive accessibility description will appear here..."
|
| 165 |
+
)
|
| 166 |
+
|
| 167 |
+
with gr.Accordion("✏️ Add New Text", open=True):
|
| 168 |
+
with gr.Row():
|
| 169 |
+
new_text = gr.Textbox(
|
| 170 |
+
label="Text to Add",
|
| 171 |
+
placeholder="Enter your text here...",
|
| 172 |
+
max_lines=3
|
| 173 |
+
)
|
| 174 |
+
|
| 175 |
+
text_position = gr.Dropdown(
|
| 176 |
+
choices=[
|
| 177 |
+
"top-left", "top-center", "top-right",
|
| 178 |
+
"center-left", "center", "center-right",
|
| 179 |
+
"bottom-left", "bottom-center", "bottom-right"
|
| 180 |
+
],
|
| 181 |
+
value="bottom-center",
|
| 182 |
+
label="Text Position"
|
| 183 |
+
)
|
| 184 |
+
|
| 185 |
+
with gr.Row():
|
| 186 |
+
font_size = gr.Slider(
|
| 187 |
+
minimum=10,
|
| 188 |
+
maximum=100,
|
| 189 |
+
value=24,
|
| 190 |
+
step=2,
|
| 191 |
+
label="Font Size"
|
| 192 |
+
)
|
| 193 |
+
|
| 194 |
+
text_color = gr.ColorPicker(
|
| 195 |
+
value="#FFFFFF",
|
| 196 |
+
label="Text Color"
|
| 197 |
+
)
|
| 198 |
+
|
| 199 |
+
background_color = gr.ColorPicker(
|
| 200 |
+
value="#000000",
|
| 201 |
+
label="Background Color",
|
| 202 |
+
info="Set to transparent for no background"
|
| 203 |
+
)
|
| 204 |
+
|
| 205 |
+
accessibility_description = gr.Textbox(
|
| 206 |
+
label="Custom Accessibility Description",
|
| 207 |
+
placeholder="Add a custom description for screen readers (optional)...",
|
| 208 |
+
max_lines=3
|
| 209 |
+
)
|
| 210 |
+
|
| 211 |
+
with gr.Row():
|
| 212 |
+
process_btn = gr.Button(
|
| 213 |
+
"✨ Add Text & Generate Accessibility Info",
|
| 214 |
+
variant="primary",
|
| 215 |
+
size="lg"
|
| 216 |
+
)
|
| 217 |
+
clear_btn = gr.Button("🗑️ Clear All", variant="secondary")
|
| 218 |
+
|
| 219 |
+
with gr.Tab("ℹ️ How to Use"):
|
| 220 |
+
gr.Markdown("""
|
| 221 |
+
### 🎯 Premium Features for Accessibility
|
| 222 |
+
|
| 223 |
+
**For Visually Impaired Users:**
|
| 224 |
+
- **Automatic Text Detection**: Our system automatically detects existing text in your uploaded images
|
| 225 |
+
- **Comprehensive Descriptions**: Get detailed accessibility information including detected text and added overlays
|
| 226 |
+
- **Screen Reader Friendly**: All outputs are optimized for screen reader compatibility
|
| 227 |
+
|
| 228 |
+
**How It Works:**
|
| 229 |
+
1. **Upload** your image using the file picker or drag-and-drop
|
| 230 |
+
2. **Enable** "Detect existing text in image" to automatically find text already present
|
| 231 |
+
3. **Add** your new text with customizable position, size, and colors
|
| 232 |
+
4. **Review** the accessibility information that combines detected text and your additions
|
| 233 |
+
5. **Download** the final image with complete accessibility metadata
|
| 234 |
+
|
| 235 |
+
**Accessibility Best Practices:**
|
| 236 |
+
- Always provide meaningful custom descriptions when possible
|
| 237 |
+
- Use high contrast colors for better visibility
|
| 238 |
+
- Consider the context of existing text when adding new overlays
|
| 239 |
+
- The accessibility information is designed to be read by screen readers
|
| 240 |
+
""")
|
| 241 |
+
|
| 242 |
+
# Event handlers
|
| 243 |
+
process_btn.click(
|
| 244 |
+
fn=add_text_to_image,
|
| 245 |
+
inputs=[
|
| 246 |
+
input_image,
|
| 247 |
+
new_text,
|
| 248 |
+
text_position,
|
| 249 |
+
font_size,
|
| 250 |
+
text_color,
|
| 251 |
+
background_color,
|
| 252 |
+
detect_existing,
|
| 253 |
+
accessibility_description
|
| 254 |
+
],
|
| 255 |
+
outputs=[
|
| 256 |
+
output_image,
|
| 257 |
+
accessibility_output,
|
| 258 |
+
existing_text_display
|
| 259 |
+
]
|
| 260 |
+
)
|
| 261 |
+
|
| 262 |
+
clear_btn.click(
|
| 263 |
+
fn=lambda: (None, "", "", ""),
|
| 264 |
+
inputs=[],
|
| 265 |
+
outputs=[
|
| 266 |
+
input_image,
|
| 267 |
+
output_image,
|
| 268 |
+
accessibility_output,
|
| 269 |
+
existing_text_display
|
| 270 |
+
]
|
| 271 |
+
)
|
| 272 |
+
|
| 273 |
+
# Auto-detect existing text when image is uploaded (if detection is enabled)
|
| 274 |
+
input_image.change(
|
| 275 |
+
fn=lambda img, detect: (detect_text_in_image(img) if detect and img is not None else []) if detect else [],
|
| 276 |
+
inputs=[input_image, detect_existing],
|
| 277 |
+
outputs=existing_text_display
|
| 278 |
+
)
|
| 279 |
+
|
| 280 |
+
# Launch the app
|
| 281 |
+
if __name__ == "__main__":
|
| 282 |
+
demo.launch()
|