Dhaksha UAV - Armed Vehicle Detection
ML system for detecting armed vehicles from micro-UAV drones. Worked with Tamil Nadu UAV Corporation and MIT Team Dhaksha for Indian Air Force's Mehar Baba Competition.
Dhaksha UAV - Armed Vehicle Detection from Micro-UAV
Overview
Developed an advanced machine learning system for detecting armed vehicles from micro-UAV drone footage as part of the Indian Air Force's Mehar Baba Competition 2. This project involved collaboration with the Tamil Nadu UAV Corporation and Team Dhaksha from MIT, working as a freelance ML engineer.
Project Context
The Indian Air Force's Mehar Baba Competition challenges teams to develop cutting-edge defense technologies. Our project focused on real-time object detection capabilities for micro-UAV drones to identify potential security threats from aerial surveillance.
Key Features
- Real-time Detection: Process drone video feeds in real-time
- Armed Vehicle Classification: Identify and classify different types of armed vehicles
- Edge Deployment: Optimized models for deployment on UAV hardware
- High Accuracy: Robust detection across varying weather and lighting conditions
- Low Latency: Critical for defense applications requiring immediate response
Technology Stack
- Deep Learning frameworks (TensorFlow/PyTorch)
- Computer Vision libraries (OpenCV)
- Python for model development and deployment
- Object detection architectures (YOLO, Faster R-CNN variants)
- Model optimization for edge devices
Impact
This project contributes to national security by enhancing surveillance capabilities and threat detection from unmanned aerial vehicles. The technology has applications in border security, military reconnaissance, and disaster response scenarios.
Achievement
Successfully shortlisted for Phase 2 of the competition among numerous participating teams from across India.