Thilagavijayan / Activity_Recognition

This project aims to build a computer vision model for activity recognition in videos. The goal is to develop a system that can accurately identify and classify various activities happening in a video, such as dancing, cleaning, cooking, shouting, fighting, theft, fall, carrying flammable objects, weapons, and baby crying etc

CV based Activity Recognition

This project demonstrates activity recognition using a pre-trained model. It takes a video as input and predicts the activity in each frame of the video.

Requirements

• Python 3.x
• OpenCV
• NumPy

Usage

To run the project using the command below:

Accessing the video frame python <file name> --model resnet-34_kinetics.onnx --classes Actions.txt --input <video path> --gpu 1 --output output.mp4

Example cmd python Activity.py --model resnet-34_kinetics.onnx --classes Actions.txt --input videos/soccer.mp4 --gpu 1 --output output.mp4

Accessing the webcam python <file name> --model resnet-34_kinetics.onnx --classes Actions.txt

Example cmd python Activity.py --model resnet-34_kinetics.onnx --classes Actions.txt