Open cestbonsuliu opened 8 months ago
This work was initially used only as the upstream unsupervised task for training pre-trained models. We did not deliberately try to obtain a complete reconstructed image with all reconstructed parts. Although I don't see why you would need to get a complete reconstructed image, if you need to do that, you do need to do multiple reconstructions and then merge them. Unfortunately we don't have the code already available, but if you encounter problems during coding, please feel free to communicate.
注意到每次推理只能重建掩码掉的图像块,需要多次推理才能获取完整的重建图像。
请问你是否有写好的代码能够获取完整的重建图像?如果有,是否能够分享呢?非常感谢!