Detect object using DETR with YOLO and a query selector with DEYO
Detect object using DETR with YOLO and a query selector with DEYO
DEYO: DETR with YOLO for End-to-End Object Detection
arXiv paper abstract https://arxiv.org/abs/2402.16370
arXiv PDF paper https://arxiv.org/pdf/2402.16370.pdf
GitHub https://github.com/ouyanghaodong/DEYO
… training … of DETRs is … contingent upon pre-training … on the ImageNet dataset. However, the limited supervisory signals provided by the image classification …
… have devised an innovative training methodology termed step-by-step training.
… in the first stage of training, … employ a classic detector, pre-trained with a one-to-many matching strategy, to initialize the backbone and neck of the end-to-end detector.
In the second stage of training, … froze the backbone and neck of the end-to-end detector, necessitating the training of the decoder from scratch.
Through … step-by-step training, … have … the first real-time end-to-end object detection model that utilizes a purely convolutional structure encoder, DETR with YOLO (DEYO).
Without reliance on any supplementary training data, DEYO surpasses all existing real-time object detectors in both speed and accuracy …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
