Zhao Xiangyu ; Chen Yicheng ; Xu Shilin ; Li Xiangtai ; Wang Xinjiang ; Li Yining ; Huang Haian

Abstract

Grounding-DINO is a state-of-the-art open-set detection model that tacklesmultiple vision tasks including Open-Vocabulary Detection (OVD), PhraseGrounding (PG), and Referring Expression Comprehension (REC). Its effectivenesshas led to its widespread adoption as a mainstream architecture for variousdownstream applications. However, despite its significance, the originalGrounding-DINO model lacks comprehensive public technical details due to theunavailability of its training code. To bridge this gap, we presentMM-Grounding-DINO, an open-source, comprehensive, and user-friendly baseline,which is built with the MMDetection toolbox. It adopts abundant vision datasetsfor pre-training and various detection and grounding datasets for fine-tuning.We give a comprehensive analysis of each reported result and detailed settingsfor reproduction. The extensive experiments on the benchmarks mentioneddemonstrate that our MM-Grounding-DINO-Tiny outperforms the Grounding-DINO-Tinybaseline. We release all our models to the research community. Codes andtrained models are released athttps://github.com/open-mmlab/mmdetection/tree/main/configs/mm_grounding_dino.