Baseline Solution exits with code 137

Bin-Picking Challenge
1

Hi,

I’m working on setting up the environment but I’m facing an issue while checking the baseline solution.
The build is completed successfully but while running

docker run --rm --name bpc_estimator  --network host   --gpus all af001c356e4d

The Non-interactive docker run fails with exit code 137.

Searching online it says that it’s due to not enough RAM but just want to make sure that if it’s a case of hardware issue or not.

For reference my laptop specs are:
16GB RAM
4070 RTX 8GB VRAM

Attaching full error code below.

Thanks!

building >  ---> Using cache
building >  ---> af001c356e4d
building > Successfully built af001c356e4d
Logging output to file ibpc_zenoh_output.log
Executing command: 
docker run --rm --name bpc_zenoh  --network host  eceadc354590 
Logging output to file ibpc_test_output.log
Executing command: 
docker run --rm -e BOP_PATH=/opt/ros/underlay/install/datasets/ -e DATASET_NAME=ipd --name bpc_tester  --network host  -v /home/{user}/bpc_ws/bpc:/opt/ros/underlay/install/datasets -v /home/{user}/bpc_ws/bpc:/submission 3c8934de816b 
Executing command: 
docker run --rm --name bpc_estimator  --network host   --gpus all af001c356e4d 
[INFO] [1742690877.589547342] [bpc_pose_estimator]: Starting bpc_pose_estimator...
[INFO] [1742690877.590043093] [bpc_pose_estimator]: Model directory set to /opt/ros/underlay/install/models.
[INFO] [1742690877.590269508] [bpc_pose_estimator]: Pose estimates can be queried over srv /get_pose_estimates.
bpc_tester finished with exit code 0 -- stopping others.
Non-interactive Docker run failed
 Command '['docker', 'run', '--rm', '--name', 'bpc_zenoh', '--network', 'host', 'eceadc354590']' returned non-zero exit status 137.
bpc_zenoh finished with exit code 137 -- stopping others.
Non-interactive Docker run failed
 Command '['docker', 'run', '--rm', '--name', 'bpc_estimator', '--network', 'host', '--gpus', 'all', 'af001c356e4d']' returned non-zero exit status 137.
Estimator finished with exit code 137
2

It could also be GPU out of memory. Are you able to run the empty solution from the main branch of bpc repo, and see the failure only on the baseline_solution branch?

3

The empty solution also exits with the same exit code.
I’ve tried monitoring my RAM and VRAM usage using nvtop and htop but it doesn’t look like it’s reaching it’s max capacity for docker to exit.

4

After testing a little bit more I can see that CPU usage goes up to 85% ~ 95% which makes me think if it’s not using the GPU?

I’m attaching a video of the issue when I run

bpc test ghcr.io/opencv/bpc/bpc_pose_estimator:example ipd
5

the empty solution doesn’t do anything, so I don’t think it can get OOM. Can you check this message thread and see if there is anything helpful there? Slack

6

Not able to join the slack channel, Could you please refresh the slack join link ?
Thanks !

7

Sorry for that, we’ll update it

8

Hi,
I am able to run the baseline solution now but only by running the components manually.

Steps I followed:
I removed all the files, packages and images downloaded to perform a clean install.

The below two commands still failed with exit code 137

bpc test ghcr.io/opencv/bpc/bpc_pose_estimator:example ipd

But building the bpc_pose_estimator and bpc_tester image and running them separately works.

docker run --init --rm --net host eclipse/zenoh:1.2.1 --no-multicast-scouting
rocker --nvidia --cuda --network=host bpc_pose_estimator:example
docker run --network=host -e BOP_PATH=/opt/ros/underlay/install/datasets -e SPLIT_TYPE=val -v<PATH_TO_DATASET>:/opt/ros/underlay/install/datasets -v<PATH_TO_OUTPUT_DIR>:/submission -it bpc_tester:latest
9

Understood, thanks for reporting this. We’ll take a deeper look. Also, the slack link should be working now.