Zhongwei Qiu
Zhongwei Qiu
After obtaining the HR images by your model, you can transmit the images into the DCT domain by cv2.dct. Then, you will get the spectral maps and you can visualize...
Thanks, we will fix it.
尝试一下用分布式启动，GPU指定为1呢？
> > 尝试一下用分布式启动，GPU指定为1呢？ > > 谢谢，已解决！把模型改成X2超分训练至iter=48k，序列长度是10，日志输出 loss: 0.0099，但是测试结果包含很多类似马赛克的效果：  请问这是正常的现象吗？我不确定是否还有必要继续训练(总的iters=400k) 我没有遇到过这个情况，或许等它再训练看看？
where is the loadddptrain.py?
> Thank you very much for your reply on the solution. We have also reproduced the relevant code ourselves and completed distributed training. Thank you again for your reply. >...