A fine-tuned YOLO model for detecting facial features and personal headwear, including glasses, hats, and masks — optimized for identity verification and KYC (Know Your Customer) applications. This model is built upon the YOLOS architecture and trained on a filtered and relabeled subset of the
Fashionpedia dataset, focusing exclusively on categories relevant to liveness verification and facial compliance checks. For more details of the implementation, check out the source code
here1CATEGORIES = [
2 'face',
3 'glasses',
4 'hat',
5 'mask',
6 'headband',
7 'head covering'
8]
