Intruder Detector
A multi-camera computer-vision surveillance prototype combining object detection, homography, image stitching, and real-world distance estimation.
September 2020 – December 2020
Built a home-surveillance prototype that used object detection to recognize people within camera coverage. I expanded the observable area by combining feeds from multiple IP cameras: each feed was transformed into a top-down view using homography with satellite imagery and then stitched into a wider scene.
I later adapted the system for social-distancing monitoring by mapping pixel distance to real-world distance and generating alerts when distancing guidelines were violated.
Areas: Computer Vision · Image Processing · Object Detection · Machine Learning · Homography · Image Stitching