Closed YSL0226 closed 8 months ago
Hi, just change the configuration files in the training and inference scripts to viton512_v2.yaml, and the ckeckpoint location used during inference also needs to be reset to the new weight file location. This update mainly uses the mask selection strategy of HR-VTON, Intuitively, the mask area is larger. Compared with the previous version, this strategy will make the generated results less likely to be affected by the shape of the person's original clothes.
Thank you for the answer
Hi authors,
Could you describe how to use cp_dataset_v2.py and what is the selection strategy update?
Looking forward to your reply