Region-centric Image-Language Pretraining for Open-Vocabulary Detection

Open paper

Benchmark	Model	Rank	Results
open-vocabulary-object-detection-on-lvis-v1-0	DITO	#2	AP novel-LVIS base training: 40.4…
open-vocabulary-object-detection-on-mscoco	DITO	#3	AP 0.5: 46.1