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

VG

Benchmarks

Task NameDataset NameSOTA ResultTrend
Multi-label recognitionVG-200
Avg OF147.9
66
Scene Graph ClassificationVG150 (test)
mR@5028.01
66
Scene Graph DetectionVG150 (test)
ng-mR@5019.8
41
Scene Graph DetectionVG150
R@5035.5
39
Predicate ClassificationVG 50 (test)
Mean Recall@5037
29
Scene Graph DetectionVG 50 (test)
mR@5022
27
Dense CaptioningVG V1.2
mAP42.8
25
Scene Graph ClassificationVG 50 (test)
R@5039.3
25
Scene Graph GenerationVG 150 (test)
R@5073.4
24
Multi-label image recognitionVG-200
Average mAP49.4
24
Scene Graph GenerationVG150
mR@5015.9
22
Predicate ClassificationVG-1800 (test)
Accuracy90.07
21
Predicate ClassificationVG150 (test)
ng-mR@5045.8
18
Scene Graph Detection (SGDet)VG-150 v1.0 (test)
R@5031
17
Scene Graph Classification (SGCls)VG-150 v1.0 (test)
R@5039.4
17
Predicate Classification (PredCls)VG-150 v1.0 (test)
R@5065.8
17
Scene Graph GenerationVG (test)
mR@5017.07
17
Predicate ClassificationVG (test)
mR@10063.02
17
Dense CaptioningVG V1.0
mAP0.424
16
Multi-label image recognition with partial labelsVG-200
mAP @ IoU=0.1040.6
15
Relation PredictionVG8K-LT v1.4 (test)
Accuracy (many)37.3
13
Relation PredictionVG200
R@5069.7
13
Multi-label RecognitionVG-200 (test)
Avg. OF145.1
13
Scene Graph DetectionVG
R@5030.26
12
Predicate ClassificationVG
Recall@5065.46
12
Showing 25 of 55 rows