An android application to classify skin lesions:
- Nevus;
- Melanoma;
- Benign keratosis;
- Basal cell carcinom;
- Actinic keratoses;
- Vascular skin lesions;
- Dermatofibroma.
preview.mov |
 |
 |
 |
|---|
EfficientNet based model trained on merged data from HAM1000 and ISIC2019. It was trained on 50 000 images (augmented data included).
| Precision | Recall | F1-score | Support | |
|---|---|---|---|---|
| akiec | 0.64 | 0.52 | 0.57 | 31 |
| bcc | 0.75 | 0.78 | 0.75 | 32 |
| bkl | 0.74 | 0.69 | 0.69 | 80 |
| df | 0.66 | 0.64 | 0.64 | 9 |
| mel | 0.48 | 0.66 | 0.55 | 44 |
| nv | 0.97 | 0.97 | 0.97 | 779 |
| vasc | 0.93 | 0.93 | 0.93 | 32 |
| -------------- | ----------- | ------------ | ---------- | --------- |
| avg | 0.75 | 0.74 | 0.74 | 957 |
| weighted avg | 0.92 | 0.91 | 0.91 | 957 |
- Recall shows if the classifier is able to detect a class;
- Precision shows if the classifier is able to detect a class correctly;
- F1 Score is a harmonic mean of the recall and precision. In common words: the higher - the better.
In case of android the model was converted to tflite with merged metadata with TensorFlow Lite metadata script.