Human Object Interaction Detection On Hico

Metrics

mAP

Results

Performance results of various models on this benchmark

Model Name	mAP	Paper Title	Repository
ACP	20.59	Detecting Human-Object Interactions with Action Co-occurrence Priors
DJ-RN	21.34	Detailed 2D-3D Joint Representation for Human-Object Interaction
TIN (TIPAMI)	20.93	Transferable Interactiveness Knowledge for Human-Object Interaction Detection
HOTR + CPC	26.16	Consistency Learning via Decoding Path Augmentation for Transformers in Human Object Interaction Detection
QPIC + CPC	29.63	Consistency Learning via Decoding Path Augmentation for Transformers in Human Object Interaction Detection
PViC-SwinL	44.32	Exploring Predicate Visual Context in Detecting Human-Object Interactions
GPNN	13.11	Learning Human-Object Interactions by Graph Parsing Neural Networks
ERNet	36.89	ERNet: Efficient and Reliable Human-Object Interaction Detection
CDN (ResNet101)	32.07	Mining the Benefits of Two-stage and One-stage HOI Detection
HOITrans(ResNet50)	23.46	End-to-End Human Object Interaction Detection with HOI Transformer
IDN (COCO detector)	23.36	HOI Analysis: Integrating and Decomposing Human-Object Interaction
ConsNet (ResNet-50)	22.15	ConsNet: Learning Consistency Graph for Zero-Shot Human-Object Interaction Detection
Ours (PViC+)	46.49	Dynamic Scene Understanding from Vision-Language Representations	-
CQL+GEN-VLKT-L	36.03	Category Query Learning for Human-Object Interaction Classification
CQL+GEN-VLKT-B	35.36	Category Query Learning for Human-Object Interaction Classification
RLIPv2 (Swin-L)	45.09	RLIPv2: Fast Scaling of Relational Language-Image Pre-training
HOITrans(ResNet101)	26.61	End-to-End Human Object Interaction Detection with HOI Transformer
GEN-VLKT-R101	34.95	GEN-VLKT: Simplify Association and Enhance Interaction Understanding for HOI Detection
Body Part Interactiveness	35.15	Mining Cross-Person Cues for Body-Part Interactiveness Learning in HOI Detection
iCAN	14.84	iCAN: Instance-Centric Attention Network for Human-Object Interaction Detection

0 of 55 row(s) selected.