DINO
DINO is the official implementation of the ICLR 2023 paper DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection. Pronounced like dinosaur, this state-of-the-art object detection model builds upon the DETR architecture by introducing improved denoising anchor boxes to accelerate convergence and boost accuracy. The software provides a complete toolkit for training and evaluating high-performance detection systems on datasets like COCO. Key capabilities include end-to-end object detection without hand-designed components, achieving leading performance metrics with efficient training schedules. Beyond the core repository, the DINO research team has expanded its ecosystem with several advanced variants. These include Stable-DINO for enhanced stability, Mask DINO for instance segmentation, Grounding DINO for zero-shot open-set detection and segmentation, Semantic-SAM for universal image segmentation, and OpenSeeD for open-set object segmentation tasks. The codebase is often integrated int