search

vt-vl-lab / 3d-photo-inpainting

[CVPR 2020] 3D Photography using Context-aware Layered Depth Inpainting
https://shihmengli.github.io/3D-Photo-Inpainting/
Other
6.91k stars 1.11k forks source link

How can I make it work on CPU only? #67

Closed d0null closed 4 years ago

d0null commented 4 years ago

I tried to change the map_location to torch.device('cpu') but it didn-t work... What are the file I need to change, can anyone help?

freeman5 commented 4 years ago

Read the documentation, you have to delete the 0 at the gpu argument in argument.yml.

d0null commented 4 years ago

Yes, I have and this happend> the process is killed

Start Running 3D_Photo ... Loading edge model at 1593091614.6673536 Loading depth model at 1593091615.3770957 Loading rgb model at 1593091617.1194363 Writing depth ply (and basically doing everything) at 1593091618.9549518 Killed

marius34130 commented 4 years ago

I had the same issue. I checked the /var/log/syslog file and it complained about not enought memory(VM with 8Gb). I will try to increase memory to 16Gb

marius34130 commented 4 years ago

I need 12GB of memory for execute the rending. Now, I have a segmentation fault. i hope it help you

d0null commented 4 years ago

I need 12GB of memory for execute the rending. Now, I have a segmentation fault. i hope it help you

Yep, I got the same error. What should be the next step?

Making video at 1593185095.3476174 fov: 53.13010235415598 Segmentation fault (core dumped)

Amenace7899 commented 4 years ago

I’m feeling your pain

On Sat, 27 Jun 2020 at 1:27 am, Daniel notifications@github.com wrote:

I need 12GB of memory for execute the rending. Now, I have a segmentation fault. i hope it help you

Yep, I got the same error. What should be the next step?

Making video at 1593185095.3476174 fov: 53.13010235415598 Segmentation fault (core dumped)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/vt-vl-lab/3d-photo-inpainting/issues/67#issuecomment-650240653, or unsubscribe https://github.com/notifications/unsubscribe-auth/APK6FSPHYBBJ6JENVNLVI6TRYS44RANCNFSM4OHMXREQ .

gagan0123 commented 4 years ago

About process getting killed(not segmentation fault), most probably its being caused by overheating of the CPU, when you're not utilizing GPU. Suggest you turn your CPU cores to a lower max frequency, then run it again. It will take a bit longer, but it will produce the results.

amatiasq commented 4 years ago

@gagan0123 Thanks for the advice. I checked it and temperature was fine. But found this issue and they suggest this command to find the reason for killed:

dmesg -T | grep -E -i -B100 'killed process'

And in my case it says

Out of memory: Killed process 45722 (python) ...

My GPU (GTX 760) is not compatible apparently and I have 7.7Gb memory, guess it's time to upgrade.