Sign language correctness discrimination (SLCD) dataset

Sign language correctness discrimination (SLCD) dataset is collected for sign language teaching. Different from general sign language recognition datasets, SLCD dataset has two kind labels of sign language category and standardization category at the same time. The standardization category is to describe action correctness of the same sign language made by students. The SLCD dataset videos in this paper are obtained by camera. 76 students are recruited to collect sign language actions. Each student collects the same gesture multiple times to ensure the diversity of the dataset. There are 52 Chinese isolate sign languages, 27 Chinese continuous sign languages. There is 20792 sign language videos in total. The videos are saved to pictures for every 2 frames. There are in total 1054598 pictures. Each saved picture has pseudo hand position labels made by semi supervised learning method. The train set and test set are divided by 9:1.