Skip to content

user1m/vott-reviewer-ext

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VOTT Reviewer Ext

Prereqs

This is a tool to run a model reviewer in a docker container, to decouple model reviewing from the VOTT application, in order to keep the application light-weight

Setup

> docker pull user1m/vott-reviewer-cntk:cpu

Usage

> git clone https://github.com/User1m/vott-reviewer-ext.git
> cd vott-reviewer-ext
> export MODEL_PATH=/absolute/path/to/cntk/model/
> ./docker/scripts/run.prod.sh cpu (or gpu)
  • NOTE: Your cntk model path must be mapped to the container's /workdir/model/ path
  • In the /workdir/model/ should be a .model file AND a class_map.txt file from your training
  • This will expose an endpoint on 127.0.0.1:3000/cntk. Plug this endpoint into VOTT and review.

Known GPU Machine Issues

NOTE - If you have a GPU machine w/ nvidia installed you might run into an issue w/ nvidia-docker & creating the container. This is due to the fact that the host already contains nvidia-cuda-toolkit binaries, found in ls -la /usr/bin/nvidia-*, and the container tries to mount it's nvidia binaries over the existing ones.

A current workaround is to rename some of the host nvidia files found in /usr/bin/nvidia-* so that the container can be created, then you can rename them back after container is created.

!! See run.prod.sh for an example of this.

Test

> curl \
  -F "image=@/home/user1/Desktop/test.jpg" \
  localhost:3000/cntk

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published