Human-Object Interaction Detection on V-COCO (Box/Mask mAP Scenarios)
72.1Box mAP (Scenario 1)RLIPv2
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| RLIPv2Backbone=Swin-L, Method Category=Additional training with object detection data2024.08 | 72.1 | 74.1 | 71.7 | 73.5 | |
| MURENBackbone=ResNet-50, Method Category=Single-stage methods2024.08 | 68.8 | 71 | 68.2 | 70.2 | |
| STIPBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 66 | 70.7 | 66.2 | 70.5 | |
| FleVRSBackbone=Focal-L, Method Category=Single-stage methods2024.08 | 65.2 | 66.5 | 66.5 | 67.9 | |
| UniVRDBackbone=ViT-L, Method Category=Additional training with object detection data2024.08 | 65.1 | 66.3 | — | — | |
| ERNetBackbone=EfficientNetV2-XL, Method Category=Single-stage methods2024.08 | 64.2 | — | — | — | |
| PVICBackbone=Swin-L, Method Category=Additional training with object detection data2024.08 | 64.1 | 70.2 | — | — | |
| CDNBackbone=ResNet-101, Method Category=Single-stage methods2024.08 | 63.9 | 65.8 | 61.3 | 63.2 | |
| GEN-VLKTBackbone=ResNet-101, Method Category=Single-stage methods2024.08 | 63.6 | 65.9 | 61.8 | 64 | |
| RLIPBackbone=ResNet-50, Method Category=Single-stage methods2024.08 | 61.9 | 64.2 | 61.3 | 64 | |
| UPTBackbone=ResNet-101, Method Category=Additional training with object detection data2024.08 | 61.3 | 67.1 | 60.3 | 65.6 | |
| QPICBackbone=ResNet-101, Method Category=Single-stage methods2024.08 | 58.3 | 60.7 | — | — | |
| DIRVBackbone=EfficientDet-d3, Method Category=Single-stage methods2024.08 | 56.1 | — | — | — | |
| HOTRBackbone=ResNet-50, Method Category=Single-stage methods2024.08 | 55.2 | 64.4 | 55 | 64.1 | |
| GGNetBackbone=Hourglass-104, Method Category=Single-stage methods2024.08 | 54.7 | — | — | — | |
| SCGBackbone=ResNet-50, Method Category=Additional training with object detection data2024.08 | 54.2 | 60.9 | 49.2 | 53.4 | |
| IDNBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 53.3 | 60.3 | — | — | |
| ACPBackbone=ResNet-152, Method Category=Bottom-up methods2024.08 | 53.2 | — | — | — | |
| FCMNetBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 53.1 | — | — | — | |
| HOI-TransformerBackbone=ResNet-101, Method Category=Single-stage methods2024.08 | 52.9 | — | — | — | |
| CHGNetBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 52.7 | — | — | — | |
| PD-NetBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 52.6 | — | — | — | |
| PMFNetBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 52 | — | — | — | |
| VSGNetBackbone=ResNet-152, Method Category=Bottom-up methods2024.08 | 51.8 | 57 | — | — | |
| DRGBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 51 | — | — | — | |
| IP-NetBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 51 | — | — | — | |
| VCLBackbone=ResNet-101, Method Category=Additional training with object detection data2024.08 | 48.3 | — | — | — | |
| TINBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 47.8 | 54.2 | — | — | |
| UnionDetBackbone=ResNet-50, Method Category=Single-stage methods2024.08 | 47.5 | 56.2 | — | — | |
| ICANBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 45.3 | 52.4 | — | — | |
| GPNNBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 44 | — | — | — | |
| InteractNetBackbone=ResNet-50, Method Category=Bottom-up methods2024.08 | 40 | — | — | — |