This repository contains the material developed by the UCB-Squirrels team for the OpenCV 2021 Spatial AI competition. It is written in Python3 and uses depthai to interact with the OAKs.
Example of gait recognition using our own dataset.
We also performed some test using the well known CASIA dataset.
- Auto-segmentation using a pretrained models
Gait recognition using the CASIA dataset
- Pedestrian detection
- Image segmentation
- Human identification
The following libraries are necessary:
- depthai (follow the instructions in the official repository)
- opencv
- tensorflow 2
- Scikit-learn
- standard libraries: matplotlib, numpy, os, glob,
- OAKD_8S dataset which is a dataset that we collected. Download it in parts part1 part2 part3 part4 main
The directories are divided in the following way:
- recording: Contains scripts to record RGB and RGB+depth data using single or multiple OAK's-D.
- recognition: Contains scripts to perform Gait recognition using appearance information.
- segmentation: Contains scripts to perform instance segmentation, specifically person/human segmentation.
- models & data: Contain necessary data for prediction
- utils: It is composed of helping scripts
We are a group of researchers and lecturers from the Bolivian Catholic University.
Picture of our team taken after recording our dataset using five OAKs
If you use this project in your research or wish to refer to the baseline results published in the README, please use the following BibTeX entry.
@misc{relabeller,
author = {UCB-Squirrels},
title = {Distributed Real-Time Gait Recognition usingOAK-D},
howpublished = {\url{https://github.com/cidimec/UCB-squirrels}},
year = {2021}
}