How to use "person re-identification" model installed in CVAT?

charlescho64 commented 4 years ago

My actions before raising this issue

[V] Read/searched the docs
[ V] Searched past issues

Expected Behaviour

In other Git for "REID model", I try to get result for searching objects similar to input object. But I don't know how to use REID model in automatic annotation functionality.

Current Behaviour

I uploaded street(including person) mp4 to cvat, and try to use reid model for automatic annotation. But I don't get any. I don't know how to use this model. I had something for other model in cvat.

Possible Solution

Let me get to use REID mode in cvat.

Steps to Reproduce (for bugs)

upload mp4 file(including person) to cvat
run automatic annotation with reid model

Context

Your Environment

Git hash commit (git log -1):
Docker version docker version (e.g. Docker 17.0.05):
Are you using Docker Swarm or Kubernetes?
Operating System and version (e.g. Linux, Windows, MacOS):
Code example or link to GitHub repo or gist to reproduce problem:
Other diagnostic information / logs:

Logs from `cvat` container

Next steps

You may join our Gitter channel for community support.

charlescho64 commented 4 years ago

Hello,

If anybody know how to use ReID in CVAT, plz let me know.

bsekachev commented 4 years ago

Hi, To install ReID model, with installed nuclio from serverless subdirectory perform:

nuctl deploy --project-name cvat \
    --path "`pwd`/openvino/omz/intel/person-reidentification-retail-300/nuclio" \
    --volume "`pwd`/openvino/common:/opt/nuclio/common" \
    --platform local

Or just run the script serverless/deploy.sh, but it will install all available models.

ReID doesn't perform person detection, it performs joining of separated shapes into tracks. For example on two frames you have the same person that is annotated by different rectangles, ReID automatically merges these shapes into the track (track is an object that exist on a range of frames, shape is an object that exist on a single frame). To person detection you can use another model (Faster RCNN for example).

omarreda14 commented 2 years ago

@bsekachev is there any tutorials or user manual to get in Re-Id ?

cvat-ai / cvat