dima806's picture
Update README.md
e26c1e9 verified
|
raw
history blame
828 Bytes
metadata
license: apache-2.0
metrics:
  - accuracy
  - f1

Returns facial emotion based on facial human image.

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

Classification report:

              precision    recall  f1-score   support

         sad     0.8324    0.8540    0.8431      3595
     disgust     0.9936    0.9981    0.9958      3595
       angry     0.8890    0.8815    0.8852      3596
     neutral     0.8626    0.8623    0.8625      3596
        fear     0.8529    0.8526    0.8527      3596
    surprise     0.9503    0.9310    0.9406      3596
       happy     0.9271    0.9260    0.9265      3596

    accuracy                         0.9008     25170
   macro avg     0.9011    0.9008    0.9009     25170
weighted avg     0.9011    0.9008    0.9009     25170