Person Detection on COCO (val)
71.1APMMPedestron
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MMPedestron#Param=62M, Evaluation Type=Finetune2024.07 | 71.1 | — | — | — | — | — | |
| MMPedestron#Param=62M, Evaluation Type=Direct2024.07 | 68.2 | — | — | — | — | — | |
| InternImage-XL#Param=387M, Protocol=Finetuned from ImageNet-22k2024.07 | 64.8 | — | — | — | — | — | |
| DINO-80#Param=218M, Training Protocol=Trained on 80 classes2024.07 | 62.3 | — | — | — | — | — | |
| DINO-Person#Param=218M, Training Protocol=Re-trained on Person category2024.07 | 61.8 | — | — | — | — | — | |
| UniHCP#Param=109M, Backbone=ViT-B, Training=Multi-task on 33 datasets2024.07 | 58.1 | — | — | — | — | — | |
| Faster RCNN-Person#Param=41M, Training Protocol=Re-trained on Person category2024.07 | 54.1 | — | — | — | — | — | |
| MultiPoseNetBackbone=ResNet-1012018.07 | 52.5 | 81.5 | 55.3 | 35.2 | 59 | 71 | |
| Faster RCNN-80#Param=42M, Training Protocol=Trained on 80 classes2024.07 | 51.9 | — | — | — | — | — | |
| MultiPoseNetBackbone=ResNet-502018.07 | 51.3 | 81.4 | 53.6 | 34.9 | 58 | 68.1 | |
| RetinaNet2018.07 | 50.2 | 77.7 | 53.5 | 31.6 | 59 | 71.5 | |
| FPN2018.07 | 47.5 | 78 | 50.7 | 28.6 | 55 | 67.4 |