Cataract Detection ResNet18 Model
Trained on ~1170 clinical images (binary: cataract vs non-cataract).
Architecture: ResNet18
Image Size: 224x224
Train Accuracy: 100%
Test Accuracy: 97.16%
Use: Baseline model for severity grading project
Future Steps
- Grad-CAM to interpret focus region (pupil/lens)
- Region-based unsupervised feature extraction
- Clustering for cataract severity grading
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support