RuntimeError: GPU out of memory about s-dcnet HOT 5 CLOSED

yeah right. Btw i'm able run your original transcript using some images from test data, it can generate the results, on WIndows. Thanks a lot. Might be due to extraction from video to images. sorry for taking your time, I am a newbie in this field. Thank you thank you.

from s-dcnet.

I debug using spyder, the error start somewhere here:

c:\users\bolt\s-dcnet\inference_wana.py(156)()
154
155
2-> 156 @hydra.main(config_path="conf/config_train_val_test.yaml")
157 def main(cfg):
3 158 orig_cwd = hydra.utils.get_original_cwd()

--Call--
KeyError: '_dh'
--Return--
KeyError: '_dh

from s-dcnet.

What are your hardware resources?

from s-dcnet.

I am using GeForce RTX 2070.

from s-dcnet.

I cannot debug everything in your code / setting.
Try launching my script inference.py for standalone image files according to the instructions in the readme. First on a Linux machine (if possible), then on Windows. If you encounter crashes likely caused by bugs in my code, please let me know.

from s-dcnet.