Typical Application Scenario
The Atlas 300I inference card (model 3000) provides an object recognition system. It uses the object detection, object tracking, object quality scoring, and high-speed object alignment algorithms to implement real-time object snapshot modeling, real-time blocklist alignment alerting, and object image retrieval.
Figure 3-1 shows the architecture of the object recognition system. The main components include the front-end HD webcams or object snapshot cameras, media stream storage server (optional), intelligent object analysis server, object alignment search server, central management server, and client management software. The Atlas 300I inference card (model 3000) is deployed in the intelligent object analysis server to implement functions such as video decoding and preprocessing, object detection inference, object alignment (correction), and object feature extraction for inference.
