double-take
double-take copied to clipboard
[BUG] ...still processing previous request
Describe the bug processing of images take long time...
Version of Double Take 1.8.0-c80a3cb
Expected behavior face-detection per image should be done within seconds.
Screenshots
Hardware virtual mashine on vmware DeepStack: Version 2022.01.01
Additional context Question: How to find the bottleneck?
i watch the logs, but how do i find the bottleneck and tweak it all to run faster? i have compreface and deepstack configured. running only one of each makes it faster, but it should also run fast with 2 detectors. whats needed? CPU? RAM? Coral? how to analyse?
looks like doubletake must first register all my collected faces in deepstack, before processing is possible. this will take a while as registering is poorly slow and i have 50+ faces to register...
i will let it do its thing and check tomorrow... luckily i am on GPU 😄 https://github.com/johnolafenwa/DeepStack/issues/127
maybe i have to downgrade deepstack? because of supported version mentioned in readme.md? will try this too...
docker pull deepquestai/deepstack:gpu-2021.06.1
🟥 not working 06.01
root@ub20-frigate4:/usr/src# docker run --gpus all -e VISION-FACE=True -v localstorage:/datastore -p 5001:5000 deepquestai/deepstack:gpu-2021.06.1
/app/server/server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.28' not found (required by /app/server/server)
i remove and deleted all dockers:
docker rm -vf $(docker ps -aq)
To delete all the images,
docker rmi -f $(docker images -aq)
Remember, you should remove all the containers before removing all the images from which those containers were created. https://stackoverflow.com/questions/44785585/docker-how-to-delete-all-local-docker-images
pulled nvida with test / and pulled deepstack again (3.2GB)
docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi
docker pull deepquestai/deepstack:gpu-2021.06.1
🟥 did not help (nvidia without error):
root@ub20-frigate4:/usr/src# docker images -a
REPOSITORY TAG IMAGE ID CREATED SIZE
deepquestai/deepstack gpu-2021.06.1 d18237a0bcdb 8 months ago 5.55GB
nvidia/cuda 11.0-base 2ec708416bb8 17 months ago 122MB
root@ub20-frigate4:/usr/src# docker run --gpus all -e VISION-FACE=True -v localstorage:/datastore -p 5001:5000 deepquestai/deepstack:gpu-2021.06.1
/app/server/server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.28' not found (required by /app/server/server)
root@ub20-frigate4:/usr/src#
🟣
docker pull deepquestai/deepstack:gpu-2021.05.31
🟥 not working
6b6a8f1005a4: Pull complete
Digest: sha256:ecf803c23a1c7999d3f8ddbd24ff69e7628226921d5223636a8624e91d96ca7b
Status: Downloaded newer image for deepquestai/deepstack:gpu-2021.05.31
docker.io/deepquestai/deepstack:gpu-2021.05.31
root@ub20-frigate4:/usr/src# docker run --gpus all -e VISION-FACE=True -v localstorage:/datastore -p 5001:5000 deepquestai/deepstack:gpu-2021.05.31
/app/server/server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.28' not found (required by /app/server/server)
❓ maybe cuda 11 vs cuda 10 ❓
🍏 test with CPU looks better, but latency increase