Quantification of grass-severing bites performed by grazing cattle using halter-mounted accelerometers and machine learning.

This repository contains various resources used for the article "Quantification of grass-severing bites performed by grazing cattle using halter-mounted accelerometers and machine learning". Below is an overview of the contents available for download.

### Contents

## 1. [Matlab_Steps_Explanation.xlsx]

- **Description:** This Excel file provides a detailed explanation of each MATLAB code used in the project. It includes information about their roles, input data requirements, and output data generated.

## 2. Data
# Base Data

- **Description:** Three ZIP files containing the raw base data used as input for the project.
- **Files:**
  - [BaseData_Behavior.zip] = The behaviors coded with the program Cowlog for each video.
  - [BaseData_Bite.zip] = The bites coded with the program Cowlog for each video.
  - [BaseIMUdata_raw.zip] = The raw signal of the IMU corresponding to each video.

# Compiled Data

- **Description:** Two ZIP files containing compiled and processed data generated during the project.
- **Files:**
  - [DataCompil_Phase1.zip] = The compilation of processed videos used for training et testing during the "Behaviour classification" phase.
  - [DataCompil_Phase2.zip]  = The compilation of processed videos used for training during the "Bite quantification" phase.

## 3. Model Tested

# Machine Learning Models

- **Description:** Two folders containing MATLAB `.mat` files of all machine learning models tested during the project.
- **Folders:**
  - [Phase1_models_tested]
  - [Phase2_models_tested]

## 4. Phase1_Code and Phase2_Code

- **Description:** Folders containing MATLAB code used in different phases of the project.
- **Folders:**
  - [`Phase1_Codes`]
  - [`Phase2_Codes`]

You can contact the author here if needed : nicolas.tilkens@junia.com

The video recordings of the animals will be put on a open access repository, the link will be updated in this description.

You can contact the author here if needed : nicolas.tilkens@junia.com