We have collected and extracted the videos from various social media sites, workout events, and Youtube videos. The videos of various Workout action classes are labelled by keeping different categories of videos in separate directories, and the directories are named in accordance with the workout action videos. After collecting the data successfully we have named our prepared data as the Workout Action Video dataset (WAVd). In the workout actions dataset the workout has been performed by various persons under different circumstances. Thus the dataset is free from biases. In our dataset, there are 11 categories of different workout actions containing 1805 videos in total. The workout action categories considered are NeckRotation, PunchingBag, HeadstandPushups, JumpingJack, PullUps, PushUps, Skipping, Squats, WallPushups, WeightLifting and GluteBridges. 

This database will only be used for Academic Research purpose, it is not for commercial use.

Some sample frame(s) of various classes present in our prepared workout action video dataset are depicted in Figure show.