Now a day’s recognizing the activity from the surveillance video is a challenging task. Some activities are not regular human activity; sometimes, it belongs to sports activity or other activities. The sports Activity is the Collection of action /activity, e.g., High jump, cricket bowling. There are different types of human activity, which are categories into three parts—first, single activity: repeating the same action in the loop. Second, Chain of available Activity: the combination of different Activities of the first type and Third, Activity interaction with object or person. Here focused on the second type of activity mostly found in sports and highlight these activities present video.
In the process of human activity recognition, there are many computer vision techniques to detect it. Many hybrid methods use two different approaches to identify activities like image processing and machine learning methods. Apply the suitable design of deep learning to fulfill the task like CNN, RNN, LSTM, and the combination of this to achieve the goal.
Example:
- High Jump : Sub Activity: Running, Jumping on a horizontal pole
- Long JUMP: Sub Activity: Running and Long Jump
- Cricket Bowling : Sub Activity: Running and Throw the Ball etc.
Mentor Name: Disha G. Deotale