Low latency object detection on the Edge-cloud AprilTag-assisted object detection and positioning

Abstract: This study proposes a low-latency video processing pipeline for object detection and positioning. The pipeline employs GPU-based inferenceframeworks and lightweight models for fast detection. Moreover, twonovel low-error pose estimation algorithms are introduced, Multi-tagsaveraging (MTA) and Multi-points embedding (MPE), which reduce estimation error to 2 cm for 4K videos. You Only Calibrate Once (YOCO)is introduced for speeding up image recovering for distorted images. The whole pipeline is flexible and can be updated with faster objectdetection models or human pose estimation models in the future. The proposed pipeline achieves a latency of 41 ms while processing 4K videos on the task of object detection and positioning.