MS
Computer VisionMachine Learning Engineer

Human Activity Recognition

A computer vision system that recognizes and classifies human activities in real time using deep learning.

PythonTensorFlowOpenCVMediaPipeNumPyScikit-learn
View Repository
Problem & Challenge

The Objective

Human activity understanding is essential for smart surveillance, healthcare monitoring, and human-computer interaction. The project demonstrates AI's ability to interpret motion patterns from live video.

Approach & Architecture

The Process

  • 1
    Collected and prepared activity datasets
  • 2
    Extracted body landmarks and motion features using MediaPipe
  • 3
    Trained deep learning models for activity classification
  • 4
    Integrated live webcam inference using OpenCV
  • 5
    Evaluated model performance and optimized prediction accuracy
Key Impact & Results

“Achieved accurate real-time activity recognition, enabling intelligent monitoring for fitness, healthcare, and surveillance applications.”