retico-team/retico-dopetrack

retico-dopetrack

A ReTiCo module for pose-tracking, based on https://github.com/naver/dope.

Installation and requirements

Please download this model and place it in the retico_dopetrack/src/models directory.

Required Packages:

• pytorch
• opencv-python
• pillow
• numpy
• scipy
• matplotlib

In addition, this module relies on retico-core and retico-vision.

See https://github.com/naver/dope for more information.

Example

import sys, os

os.environ['PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION'] = 'python'

prefix = '/path/to/prefix/'
sys.path.append(prefix+'retico-core')
sys.path.append(prefix+'retico-vision')
sys.path.append(prefix+'retico-dopetrack')

from retico_vision.vision import WebcamModule 
from retico_core.debug import DebugModule
from retico_dopetrack.dopetrack import DopeTrackingModule

cam = WebcamModule()
dope = DopeTrackingModule(visualizer=True)
debug = DebugModule(print_payload_only=True)

cam.subscribe(dope)
dope.subscribe(debug)

cam.run()
dope.run()
debug.run()

input()

debug.stop()
dope.stop()
cam.stop()

Citation

The source code this module is based on comes from https://github.com/naver/dope, which is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 license. Minor changes have been made to facilitate better interfacing. The DOPE model is a result of the following ECCV'20 paper:

@inproceedings{dope,
  title={{DOPE: Distillation Of Part Experts for whole-body 3D pose estimation in the wild}},
  author={{Weinzaepfel, Philippe and Br\'egier, Romain and Combaluzier, Hadrien and Leroy, Vincent and Rogez, Gr\'egory},
  booktitle={{ECCV}},
  year={2020}
}