Using object classification to recognise hand signs and decode sign language. Will be using a normal CNN with softmax output layer. Preprocessing will include detecting the hand and cropping the image to make classification easier. The dataset can be that of American Sign Language, or Indian Sign Language, with the former being more abundant.
ASL Sign language dataset (kaggle)
- 87k files arranged character-wise (3000 each, for 29 characters, including space, delete and nothing)
- .jpg file format
- Vegetable recognition and using it to generate recipes using provided vegetables
- Detecting people using yolo, to check violation of social distancing in public areas