JTA Human Key Point Tracking Dataset

JTA (Joint Track Auto) is a human key point tracking dataset. This dataset uses a realistic video game to track human key points in urban scenes. The dataset contains about 500,000 frames and nearly 10 million body poses.