ResearchBenchmarksBounding box classification on BoxClass-CN566Follow68.6Top-1 AccuracyFG-CLIP 254.66458.28261.965.518Oct 13, 2025Evaluation ResultsMethodMethodLinksTop-1 AccuracyFG-CLIP 2Backbone=ViT-L/16Backbone=ViT-L/162025.1068.6FG-CLIP 2Backbone=ViT-So/16Backbone=ViT-So/162025.1066.5SigLIP 2Backbone=ViT-So/16Backbone=ViT-So/162025.1063.6FG-CLIP 2Backbone=ViT-B/16Backbone=ViT-B/162025.1060.7SigLIP 2Backbone=ViT-B/16Backbone=ViT-B/162025.1057.9SigLIP 2Backbone=ViT-L/16Backbone=ViT-L/162025.1056.6Meta CLIP 2Backbone=ViT-H/14Backbone=ViT-H/142025.1055.2