Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Reasoning Segmentation

Benchmarks

Task NameDataset NameSOTA ResultTrend
Multiple Object GroundingReasoning Segmentation Overall
gIoU (%)48.7
10
Multiple Object GroundingReasoning Segmentation Long query
gIoU49.2
10
Multiple Object GroundingReasoning Segmentation Short query
gIoU47.1
10
Reasoning SegmentationReasoning Segmentation augmented (val)
cIoU51.43
3
Object Existence DetectionReasoning Segmentation augmented (val)
FP Recall86.5
3
Showing 5 of 5 rows