How karnabhavishya/cnn_melonama-detection works
This project classifies uploaded or selected skin-lesion images as benign or malignant using a trained CNN. The web workflow accepts an image, stores and preprocesses it, runs inference, then renders a diagnosis with confidence and chart data; a separate Tkinter desktop workflow performs image selection and inference. Training builds the CNN from labeled dataset folders and saves its model for both interfaces. The README also describes training-history metrics, though the sampled web code uses hard-coded history values. Template and chart wiring is documented; unsampled template behavior is not inferred beyond that.
Main entry points: app.py Flask app · train.py Training pipeline · predict.py Tkinter app · cnn_skin_lesion_high_acc.h5 Trained weights
Components 14
Web diagnosis
- Flask app
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/app.py - Uploaded image
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/static/uploads - CNN model
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/cnn_skin_lesion_high_acc.h5 - Result dashboard
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/templates/result.html - Upload page
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/templates/index.html - Chart assets
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/static/charts
Desktop diagnosis
- Tkinter app
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/predict.py - CNN model
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/cnn_skin_lesion_high_acc.h5
Model training
- Training pipeline
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/train.py - Labeled images
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/dataset - Trained weights
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/cnn_skin_lesion_high_acc.h5 - History log
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/data/history.csv
Other components
- User
- Desktop user
Connections 13
- User → Flask app: uploads image
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/app.py - Flask app → Upload page: renders
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/app.py - Flask app → Uploaded image: saves image
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/app.py - Flask app → CNN model: predicts with
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/app.py - Flask app → Result dashboard: renders diagnosis
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/app.py - Flask app → Chart assets: serves assets
README.md - Desktop user → Tkinter app: selects image
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/predict.py - Tkinter app → CNN model: predicts with
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/predict.py - Training pipeline → Labeled images: trains on
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/train.py - Training pipeline → Trained weights: saves weights
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/train.py - Training pipeline → History log: records metrics
- Flask app → Trained weights: loads weights
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/app.py - Tkinter app → Trained weights: loads weights
Final Year Project Batch-14(CNN-Based Melanoma Detection from Skin Lesion Images)/predict.py