TrackNet is a dataset for identifying tennis moves in sports.

The first dataset is based on the broadcast video of the men's singles tennis final of the 2017 Summer Universiade. The resolution, frame rate, and video length are 1280×720, 30 fps, and 75 minutes, respectively. After deletion, 81 game-related video clips were obtained, each of which recorded the complete process from the start of the serve to the score. There are 20,844 labeled frames in the first dataset.

The second dataset consists of clips from 9 broadcast videos. Each video has about 2000 frames. The second dataset includes 23,903 labeled frames.

The second dataset, along with a subset of the first dataset, was used for cross-validation and training the general model.