Docker env + working demo video + docker images link

[XGenietony]

I don't know what is going on. I just try to test python -m pyk4a.viewer --undistort_depth --vis_depth --parallel_bt I have connected my Kinect Camera. But it still can not run. The Bugs look like these:

[rmbashirov]

Hi!

Glad you managed to install our docker.

I've tested current code on a new machine today and it worked.

Can you make an additional screenshot of running ./run_local.sh and nvidia-smi after that?

[rmbashirov]

@XGenietony Does sudo xhost + from host (not from docker container) work?

[XGenietony]

Hi!

Glad you managed to install our docker.

I've tested current code on a new machine today and it worked.

Can you make an additional screenshot of running ./run_local.sh and nvidia-smi after that?

Actually i have solved this problem. I think we need to change the shell ./run_local.sh like this:

#!/usr/bin/env bash

source source.sh

VOLUMES="-v /home/xgenie:/home/xgenie -v /storage:/storage -v $PWD/../src:/src"

# ensure nvidia is your default runtime
# docker run -it --rm --name test --gpus all ubuntu:latest (it may not pull the GPU up)
sudo docker run -it --rm $PARAMS $VOLUMES --gpus all $NAME_03 $@

Because the original shell can not use the Nvidia GPU, it caused OpenGL problem.

For another, i think the ./run_server should be changed like this:

#!/usr/bin/env bash

python server.py --config ./config/server/renat.yaml

We can direct run server.py in docker.

[XGenietony]

Thank your code & reply. I just met problem like #2 . I search this bug. The version of Numpy may <=1.19.5. But i found Numpy 1.20.2 in my docker. So i think we need to change the Numpy Version in docker.

[rmbashirov]

@XGenietony

You're right, the default version of numpy changed, I checked only pyk4a.viewer yesterday.

I've updated the code 5e7560d27e70f302ba25aeebac7581b569e25fe6. I also contains instructions on how to run docker without sudo. Whole system works.

I guess you've managed to run our system, can you share a representative picture of our system running?

[XGenietony]

@XGenietony

You're right, the default version of numpy changed, I checked only pyk4a.viewer yesterday.

I've updated the code 5e7560d. I also contains instructions on how to run docker without sudo. Whole system works.

I guess you've managed to run our system, can you share a representative picture of our system running?

Sorry, because of Numpy problem, i can not completely run your system. Now, I can run your system without hand pose predict. I am updating my docker. When i can completely run the system, i will upload my sreenshot.

PS: we may save the docker images as tar files and upload. The other users can recover docker images from tar files. It will help people quick start. :)

[rmbashirov]

@XGenietony Rebuilding with new numpy will take only 10 mins.

From what I see, the problem is not to build image, but to install docker ;(

[XGenietony]

@XGenietony Rebuilding with new numpy will take only 10 mins.

From what I see, the problem is not to build image, but to install docker ;(

Sorry, i change a new machine and my colleagues use my camera for another project, so i can not upload my screenshot immediately.

Because of chinese firewall, rebuilding docker is so difficult. But, i still update my docker with Numpy 1.19.5. I upload the tar file on :https://drive.google.com/file/d/1h9TTmysSfBCzRx9nqwQ-LF4zkZuSKF4-/view?usp=sharing For Chinese developers https://pan.baidu.com/s/1yYgFSHO0O46HzimFf1PMKg Password vu1u

It might help other developers.

[rmbashirov]

@XGenietony Thank you

I'll check next week, If all good, I'll add both links to main readme.

[XGenietony]

Hey bro. , i take my camera back, and this is the demo

https://user-images.githubusercontent.com/42089807/127603735-f68f05ae-8b00-410c-b8a7-cdb4cdfaadb4.mp4

[rmbashirov]

Baidu link works! See link in comment from @XGenietony Download azure_kinect.tar Do docker load < azure_kinect.tar You will see smth like Loaded image ID: sha256:7f93c908099c14cee0baff865f307b464ef66c8631174969c8eb812ba62f4a5d Then do docker tag sha256:7f93c908099c14cee0baff865f307b464ef66c8631174969c8eb812ba62f4a5d azure_kinect:03_final

[rmbashirov]

@XGenietony I updated master readme.md 03180723c99759ba2500bcd42b5fe7a1d26eb507 Cannot access google drive link due to permission request, can you make public link?

[ykk648]

reminds 2 things:

1. the host needs kinect driver ~2. the host needs conda env with requirements~
2. failed on a cuda11&rtx3080 pc, succeed on a cuda10.2&1080 pc