Returns hand gesture based on image with about 96% accuracy.
See https://www.kaggle.com/code/dima806/hand-gestures-image-detection-vit for more details.
Classification report:
precision recall f1-score support
call 0.9261 0.9712 0.9481 11826
dislike 0.9773 0.9840 0.9807 11825
fist 0.9839 0.9871 0.9855 11825
four 0.9150 0.9326 0.9237 11825
like 0.9743 0.9110 0.9416 11825
mute 0.9835 0.9973 0.9903 11826
ok 0.9579 0.9657 0.9618 11826
one 0.9681 0.9467 0.9573 11826
palm 0.9781 0.9618 0.9699 11826
peace 0.9205 0.9323 0.9263 11825
peace_inverted 0.9777 0.9751 0.9764 11826
rock 0.9428 0.9387 0.9408 11825
stop 0.9522 0.9735 0.9627 11826
stop_inverted 0.9825 0.9561 0.9691 11826
three 0.9105 0.9114 0.9110 11826
three2 0.9800 0.9671 0.9735 11826
two_up 0.9544 0.9747 0.9644 11825
two_up_inverted 0.9762 0.9697 0.9730 11826
accuracy 0.9587 212861
macro avg 0.9590 0.9587 0.9587 212861
weighted avg 0.9590 0.9587 0.9587 212861
- Downloads last month
- 191