This is the code for my ICPR 2024 paper titled ["Hand over face gesture classification with Feature driven vision transformer and supervised contrastive learning"]. I hope you find this useful
The proposed algorithm in the paper ["Hand over face gesture classification with Feature driven vision transformer and supervised contrastive learning"] has two major steps. The end goal of our algorithm is to perform hand over face (HOF) gesture recognition. First we perform a contrastive learning using TResNet-L with supervised contrastive loss. In the second step we we use our proposed model feature driven transformer with the pre-trained TResNet encoder for final training on classification. Please see the figure for complete model understanding.