dinov2_vitb14 or resnet ？

[supercaptain666]

Is encoder_sample used in the code resnet or dinov2_vitb14? If it is dinov2_vitb14, paddingsize14 cannot be divisible by imagesize256? Is there a problem?

[Luckypuppyyjl]

1.You can swap out scape/models/detectors/ for transformer_pose.py, which uses ResNet as the backbone. 2.When the backbone is set to dinov2_vitb14, the image size should be 224. You can modify this in the configuration file (just the image size)

[supercaptain666]

ooo，thanks