Add bias correction for age predictions - Model was over-predicting ages due to aggressive sample weighting - Apply correction formula to return realistic ages - Show both corrected and raw predictions for transparency

CRITICAL FIX: Correct preprocessing normalization mismatch - Replace ResNet50 preprocess_input with simple /255.0 normalization - Fixes negative age predictions

Fix Gradio output mismatch: return tuple instead of dict for two Number components

Upload app.py with huggingface_hub