trying to run in inference mode about flownet2-pytorch

[0asa]

Hi!

Check the inference_dataset_root directory. I ran into the same issue, it turned out the path was not accessible/correct. In your example, your directory is /../frame_00, it is looking weird, especially the first /. Also, if you are using the docker image, make sure to mount the volume accordingly so your image folder is accessible from within the docker image.

[fitsumreda]

@bkvie @coderclear this happens when the inference_dataset_root path doesn't exist.

[fitsumreda]

yes, this likely because the inference_dataset_root point to an empty folder or it doesn't exist

[coderclear]

I also met this issue. Have you solved?In 251 line of main.py define the 'inference_loader'.

[coderclear]

if inference_root is 'main.py --inference --model FlowNet2 --save_flow --inference_dataset MpiSintelClean --inference_dataset_root /../frame_00 --resume ./checkpoints/flownet2'.
i got
NameError: name 'inference_loader' is not defined
if inference_root is 'main.py --inference --model FlowNet2 --save_flow --inference_dataset MpiSintelClean --inference_dataset_root /../training/clean/alley_1 --resume ./checkpoints/flownet2'.
I got
self.frame_size = frame_utils.read_gen(self.image_list[0][0]).shape
IndexError: list index out of range
and
self.size type is <class 'int'>
self.size is 0
[0.000s] Operation failed
@0asa

[Estrella1024]

Have you solved this problem?

[Estrella1024]

我也遇到了这个问题。解决了吗？在main.py的251行中定义了'inference_loader'。

How do you define it?

[Sidd1609]

@fitsumreda I am having this issue despite having a non-empty & existing folder with images that have shape in multiples of 64.

[coderclear]

已收到！

[ruffiann]

@fitsumreda I am having this issue despite having a non-empty & existing folder with images that have shape in multiples of 64.

I've noticed that when I process .jpg files it reports a 'list index out of range' error, but works fine when processing .png files. You can try it, hope it helps!

[coderclear]

已收到！