Re-identification (Re-ID) builds an appearance fingerprint so a subject seen on one camera or before an occlusion can be matched when it reappears. It keeps IDs consistent across blind spots and multiple cameras, but raises the same privacy questions as recognition.
The book · Volume 6 of 7
AI for Video: A Complete Guide to Intelligence in Video Products: From Models and Inference to Computer Vision, Speech, Search, and Generative Video
Re-identification is where multi-camera tracking usually breaks: IDs swap after occlusions or when a person walks between cameras. Volume 6 covers re-identification in detail as part of computer vision: detection, multi-object tracking, appearance embeddings and the deployment topologies that decide whether matching runs at the edge or in the cloud. Written by Nikolay Sapunov, CEO at Fora Soft.
Losing track IDs when people move between cameras or behind occlusions?
Share a sample of your camera layout and footage, and we will propose a re-identification setup with expected accuracy and hardware needs. 250+ video products shipped since 2005.
