Mice Tracking Using The YOLO Algorithm

Output examples of the YOLO network. (a)-(c) refer to Ethological Evaluation, (d)-(f) refer to Automated home-cage and (g)-(i) to Crim

This project was developed and tested for Ubuntu 16.04 and 18.04.

Knowledge in the darknet framework and YOLO object detector is required, further reading can be found here.

The darknet framework present here is a reimplementation based in pjreddie and AlexeyAB work.

Requirements

• Nvidia Drivers 390.87: Install Guide

• CUDA 9.1: Install Guide

• CuDNN 7.1: Install Guide

• GCC/G++ 5.5.0: Install Guide

• OpenCV 3.3.0: Install Guide

Installing Darknet

First clone the repository and complie the source code. This can be accomplished by:

user@computer:~$ git clone https://gitlab.com/helton.maia/proj-cnn-mice
user@computer:~$ cd proj-cnn-mice
user@computer:~/proj-cnn-mice$ make

Weights

The latest weights used in the work can be downloaded here: Full YOLO, Tiny YOLLO.

Using the object detector

Clone the repository and enter the folder or add the darknet executabel to the path variable.

• To classify a single image do:

user@computer:~/repo_folder$ sh image.sh .data .cfg .weight image

• To classify a video do:

user@computer:~/repo_folder$ sh video.sh .data .cfg .weight video

• To save a video with the classifications do:

user@computer:~/repo_folder$ sh videoOutput.sh .data .cfg .weight video output

• To calculate the mAP of a weight do:

user@computer:~/repo_folder$ sh mAP.sh .data .cfg .weight

• To get the execution time spent in a file do:

user@computer:~/repo_folder$ sh executionTime.sh .data .cfg .weights files output

Cite This Project

If you use this project in your research or wish to refer to the baseline results published, please use the following BibTeX entry.

@article{peixoto2019mice,
  title={Mice tracking using the YOLO algorithm},
  author={
    Peixoto, Helton Maia and Teles, Richardson Santiago and Luiz, John Victor Alves and Henriques-Alves, Aron Miranda, and Santa Cruz, Rossana Moreno
  },
  journal={PeerJ Preprints},
  volume={7},
  pages={e27880v1},
  year={2019},
  publisher={PeerJ Inc. San Diego, USA}
}