LesNet (Lesion Net) is an open-source project for AI-based skin lesion detection. It aims to create a reliable tool and foster community involvement in critical AI problems. Contributions are welcome!
🎨 📝 🐛 ⚡ 🔧 Update styles, rename project, remove bad augmentations, add hparam tuner, and update frontend
🎨 Change the style of the buttons so they don't overlap each other and other elements.
📝 Rename project to LesNet (Lesion Net).
🐛 Remove bad augmentations that affect training data.
⚡ 🔧 Add hyperparameter tuner (experiments).
🎨 Update frontend with a new look and feel.
⚡ 🔧 Update batch size and learning rate configurations, add data augmenter and improve inference. Changes include:
Setting batch size to 256.
Updating starting learning rates to 0.001.
Adding a data augmenter that does not require downloading all images.
Making inference and label improvements.
Introducing a new model quantize command.
Directory cleanup.
adjusting learning rate cooldown to 5 and lr patience to 7 and early stopping to 42 - more aggressive training
🎨 📝 🐛 ⚡ 🔧 Update styles, rename project, remove bad augmentations, add hparam tuner, and update frontend
🎨 Change the style of the buttons so they don't overlap each other and other elements. 📝 Rename project to LesNet (Lesion Net). 🐛 Remove bad augmentations that affect training data. ⚡ 🔧 Add hyperparameter tuner (experiments). 🎨 Update frontend with a new look and feel. ⚡ 🔧 Update batch size and learning rate configurations, add data augmenter and improve inference. Changes include: