Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

ARMBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Object Instance SegmentationARMBench Mixed-Object Tote (test)
mAP5086.37
44
Object IdentificationARMBench (test)
Recall@198
10
Reward ModelingARMBench-VL ours (test)
FG Score67.6
7
Failure Detection and ReasoningARMBench
Detect Acc.65
6
Post-stow bin state predictionARMBench Bin Sweep instance-mask space (test)
N-IoU64.22
4
Post-stow bin state predictionARMBench Direct Insert instance-mask space (test)
N-IoU70.21
4
Defect DetectionARMBench
Multi-Pick Precision84
3
Failure Detection and ReasoningARMBench S→A
Detection Accuracy72.5
2
Object Instance SegmentationARMBench Same-Object Tote (test)
mAP5015
2
Object Instance SegmentationARMBench Zoomed-Out Tote (test)
mAP5057
2
Showing 10 of 10 rows