DJW cdece0b32a 第一次提交 | 9 mēneši atpakaļ | |
---|---|---|
.. | ||
animalpose | 9 mēneši atpakaļ | |
ap10k | 9 mēneši atpakaļ | |
locust | 9 mēneši atpakaļ | |
zebra | 9 mēneši atpakaļ | |
README.md | 9 mēneši atpakaļ |
Top-down methods divide the task into two stages: object detection, followed by single-object pose estimation given object bounding boxes Instead of estimating keypoint coordinates directly, the pose estimator will produce heatmaps which represent the likelihood of being a keypoint, following the paradigm introduced in Simple Baselines for Human Pose Estimation and Tracking.
Results on AnimalPose validation set (1117 instances)
Model | Input Size | AP | AR | Details and Download |
---|---|---|---|---|
HRNet-w32 | 256x256 | 0.740 | 0.780 | hrnet_animalpose.md |
HRNet-w48 | 256x256 | 0.738 | 0.778 | hrnet_animalpose.md |
ResNet-152 | 256x256 | 0.704 | 0.748 | resnet_animalpose.md |
ResNet-101 | 256x256 | 0.696 | 0.736 | resnet_animalpose.md |
ResNet-50 | 256x256 | 0.691 | 0.736 | resnet_animalpose.md |
Results on AP-10K validation set
Model | Input Size | AP | Details and Download |
---|---|---|---|
HRNet-w48 | 256x256 | 0.728 | hrnet_ap10k.md |
HRNet-w32 | 256x256 | 0.722 | hrnet_ap10k.md |
ResNet-101 | 256x256 | 0.681 | resnet_ap10k.md |
ResNet-50 | 256x256 | 0.680 | resnet_ap10k.md |
CSPNeXt-m | 256x256 | 0.703 | cspnext_udp_ap10k.md |
Results on Desert Locust test set
Model | Input Size | AUC | EPE | Details and Download |
---|---|---|---|---|
ResNet-152 | 160x160 | 0.925 | 1.49 | resnet_locust.md |
ResNet-101 | 160x160 | 0.907 | 2.03 | resnet_locust.md |
ResNet-50 | 160x160 | 0.900 | 2.27 | resnet_locust.md |
Results on Grévy’s Zebra test set
Model | Input Size | AUC | EPE | Details and Download |
---|---|---|---|---|
ResNet-152 | 160x160 | 0.921 | 1.67 | resnet_zebra.md |
ResNet-101 | 160x160 | 0.915 | 1.83 | resnet_zebra.md |
ResNet-50 | 160x160 | 0.914 | 1.87 | resnet_zebra.md |