Stereo Drone Detection System

A sophisticated two-camera voxel-based drone detection system using ray intersection voting. Uses two iPhone X cameras positioned ~165 meters apart to detect moving objects (drones) in the sky. The system casts rays from 2D blob detections into a shared 3D voxel grid (100x150x100m at 1m resolution) and finds convergence points.

By leveraging stereo vision principles with a wide baseline, the system can accurately triangulate drone positions in 3D space. Frame differencing extracts moving blobs, which are then projected as rays into a volumetric grid. Where rays from both cameras intersect, a drone is detected with full 3D coordinates, enabling trajectory tracking and real-time CSV output for further analysis.

Key Features

• Dual-camera stereo detection (~165m baseline)
• Voxel-based 3D ray intersection voting
• Frame differencing and blob extraction
• 2D tracking with 3D track persistence filtering
• Real-time trajectory CSV output
• Annotated detection videos with 3D coordinates
• 3D trajectory visualization

Demo Video