res-loglikelihood-regression
res-loglikelihood-regression copied to clipboard
validate.py报错
我用了您提供的与训练模型,在coco val2017上测试模型的性能,但是报出如下错误(python3.6 torch1.8):

The error message is very clear, the shape of the input is wrong. As the input for pytorch model, the shape should be [batch_size, 3, 256, 192], but you got [3, 256, 192]. The input variable "inps" is generated by gt_val_loader, check it, I guess you might change some codes.
I have also encountered this problem. Have you resolved it