Hello, I tried running your code. I have installed LUA Torch, Cuda, etc, everything needed inside a Ubuntu docker. I have downloaded the pretrained model and placed it in the right folder (SCNN/experiments/pretrained/vgg_SCNN_DULR_w9.t7). There seems to be no error that I can see. It starts processing images, 0.223 seconds per image on average, and it goes through all the images.
The problem is, all of the images are completely black on the output, and of same filesize (only 303 bytes), and all output text files are simply "0 0 0 0". It feels like the algorithm is simply not doing anything at all. Have you maybe had this issue and/or know of a solution?
Hello, I tried running your code. I have installed LUA Torch, Cuda, etc, everything needed inside a Ubuntu docker. I have downloaded the pretrained model and placed it in the right folder (
SCNN/experiments/pretrained/vgg_SCNN_DULR_w9.t7). There seems to be no error that I can see. It starts processing images, 0.223 seconds per image on average, and it goes through all the images.The problem is, all of the images are completely black on the output, and of same filesize (only 303 bytes), and all output text files are simply "0 0 0 0". It feels like the algorithm is simply not doing anything at all. Have you maybe had this issue and/or know of a solution?