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

CrossTask

Benchmarks

Task NameDataset NameSOTA ResultTrend
Procedure PlanningCrossTask
Success Rate (SR)40.45
43
Goal-conditioned visual planningCrossTask T=4 88 (test)
SR37.04
40
Action Step LocalizationCrossTask (test)
Recall52.5
32
Action step localizationCrossTask
Average Recall47.3
28
Goal-conditioned visual planningCrossTask T=3 88
Success Rate (SR)47.47
27
Procedure PlanningCrossTask T=3 (test)
SR41.14
27
Visual PlanningCrossTask
Success Rate (SR)38.45
22
Online Action DetectionCrossTask
P-F134.5
20
Keystep recognitionCrossTask (test)
Accuracy28.9
18
Keystep recognitionCrossTask
Accuracy64.5
17
Procedure PlanningCrossTask T=5
Success Rate14.2
15
Goal-conditioned visual planningCrossTask T=3 88 (test)
Success Rate (SR)51.71
13
Procedure LearningCrossTask
Precision60.9
13
Consistent Video RetrievalCrossTask (test)
Accuracy0.6436
13
Keystep forecastingCrossTask
Accuracy30.2
12
Task recognitionCrossTask
Accuracy97.1
12
Procedure PlanningCrossTask short horizon T=3
SR37.96
11
Weakly-supervised Action SegmentationCrossTask
MoF54
11
Procedure PlanningCrossTask short horizon T=4
SR22.56
10
Procedure PlanningCrossTask long horizons T=6
Success Rate (SR)9.27
10
Action SegmentationCrossTask
F1 Score61.4
9
Temporal Action LocalizationCrossTask
Recall41.4
9
Temporal Action LocalizationCrossTask (test)
Recall0.414
9
Procedure PlanningCrossTask T=4 (test)
SR16.41
8
Step localizationCrossTask
Recall49.7
8
Showing 25 of 41 rows