search

NVIDIA / semantic-segmentation

Nvidia Semantic Segmentation monorepo
BSD 3-Clause "New" or "Revised" License
1.76k stars 388 forks source link

Fail to run inference and dump images on a folder of images #138

Closed xinghaoxinghao closed 3 years ago

xinghaoxinghao commented 3 years ago

when I try to run inference and dump images on a folder of images,the output is shown . image

karansapra commented 3 years ago

Hmm, this is weird its not able to detect train.py, can you check train.py exists in the location above (rough bird ) , please.

ajtao commented 3 years ago

Hello, the way the scripts here work is that they first try to create a new "run" directory under LOGROOT, which is defined in your .runx file. All the code is copied there, and then the training job is run from within that "run" directory. The idea is that this preserves the state of the code and also saves all the log output into a single "run" directory tree, so you can refer back to this saved state in the future, if needed.

In your case, the "run" directory is logs/dump_folder/rough-bird_.... If the directory logs doesn't exist, then it would have this complaint that you're seeing.

xinghaoxinghao commented 3 years ago

Thank you.However，the directory 'logs' have existed, and the code can copy sucessfully when i run inference and dump images on a folder of images. image

desti-nation commented 2 years ago

Thank you.However，the directory 'logs' have existed, and the code can copy sucessfully when i run inference and dump images on a folder of images. image

have you addressed this problem? I meet the save question.