Detecting Human-Object Interactions with Object-Guided Cross-Modal Calibrated Semantics

Benchmark Model Rank Results
human-object-interaction-detection-on-v-cocoOCN (ResNet101)#7AP(S1): 65.3AP(S2): 67.1
human-object-interaction-detection-on-v-cocoOCN (ResNet50)#9AP(S1): 64.2AP(S2): 66.3Time Per Frame(ms): 43
human-object-interaction-detection-on-hico-detOCN (ResNet101)#28mAP: 31.43