dima806's picture
Update README.md
d58c7ee verified
metadata
license: apache-2.0
metrics:
  - accuracy
  - f1
base_model:
  - google/vit-base-patch16-224-in21k

Returns face obstruction type given a facial image with about 91% accuracy.

See https://www.kaggle.com/code/dima806/face-obstruction-image-detection-vit for more details.

Classification report:

              precision    recall  f1-score   support

  sunglasses     0.9974    0.9985    0.9980      3422
     glasses     0.9896    0.9968    0.9932      3422
       other     0.7198    0.7613    0.7400      3422
        mask     0.9971    0.9985    0.9978      3422
        hand     0.7505    0.7086    0.7290      3422
        none     0.9976    0.9860    0.9918      3422

    accuracy                         0.9083     20532
   macro avg     0.9087    0.9083    0.9083     20532
weighted avg     0.9087    0.9083    0.9083     20532