QueenBeeDetection

This is the source code and implementation of a method using traditional machine learning in order to autonomously detect and recognize a bee queens and workers in a beehive. It has been used in the context of an international conference.

---

Table of Contents

1. System Overview
2. Built With
3. Installation
4. Basic Usage
5. Features
6. Contributing
7. References
8. License

---

System Overview

The proposed sysem might be described by the following figure:

It consists of three main components namely:

• Feature extraction with HOG (Histogram of Oriented Gradients),
• Dimensionality reduction with PCA (Principal Components Analysis),
• Classification with SVM (Support Vector Machines).

Built With

This project is built using these technologies:

• Programming Language: Python 3.7.6 x64
• Libraries:
  • Numpy: https://numpy.org/
  • OpenCV: https://opencv.org/
  • Scikit-Image: https://scikit-image.org/
  • Scikit-Learn: https://scikit-learn.org/
• Integrated Development Environment: Microsoft Visual Studio code x64

Installation

In order to use this implementation, one has to proceed as follows:

1. Install pip for package management and virtualenv for virtual environnements,
2. Create a virtual environnement with virtualenv and activate it : virtualenv venv,
3. Install all the dependecies using the provided requirement.txt : python -m pip install -r requirements.txt,
4. Use the run.py with the instructions below (see Basic Usage).

Basic Usage

At any time, you may execute the command python ./run.py -h to display the help & instructions:

usage: run.py [-h] [-t] [-f] [-c] [-r] [-i] [-e] [-df]

Queen Bee Detection and Recognition by Yacine YADDADEN [ https://github.com/yyaddaden ]

optional arguments:
  -h, --help           show this help message and exit

train a model:
  -t, --train          training
  -f , --folder        traning folder
  -c , --components    number of components

perform recognition:
  -r, --recognition    recognition
  -i , --image         bee image

model evaluation:
  -e, --evaluation     evaluation
  -df , --datafolder   dataset folder

1. Training a model

The objective of this operation is to generated a trained model using the images from a specific folder containing twi distinct sub-folders (queen and worker). In our case, the folder dataset provided in this repository might be used.

The command to use consists in: python run.py -t -f "dataset/" -c 25

Where "dataset/" represents the data used for the learning phase and 25 the number of principal component.

This operation will generates two model files: svm.csv and pca.csv.

2. Detecting/recognizing a bee queen

The objective of this operation is to test the prediction capability of the generated model by feeing it with an input image.

The command to use consists in: python run.py -r -i "dataset/queen/002.png"

3. Evaluating a model

The objective of this operation is to assess the peroformance of the proposed method. It will use the 10-folds cross-validation strategy during the evlauation.

The command to use consists in: python run.py -e -df "dataset/"

It will find automatically the best number of principal components to use.

It also evaluates on the basis of two main criteria namely: accuracy and confusion matrix.

Features

For now, there are three main features which consist in :

1. Training a model using the provided dataset with a specific number of principal components,
2. Testing the generated model by the detection of the bee queen presence in an image,
3. Evaluating the performance of the model using the 10-folds cross-validation strategy.

Contributing

In order to contribute to this project, there are two options :

• Option 1 : 🍴 Fork this repo!
• Option 2 : 👯 Clone this repo to your local machine using https://github.com/yyaddaden/QueenBeeDetection.git

References

In order to use the following source code or bee dataset, please make sure to cite the following paper:

Marquis, M., Yaddaden, Y., Adda, M., Gingras, G., & Coriveau-Côté, M. (2021). Automatic Honey Bee Queen Presence Detection on Beehive Frames Using Machine Learning. In The 11th International Conference on Robotics, Vision, Signal Processing, and Power Applications (RoViSP) (pp. 1-6). Springer.

In BiBtex:

@InProceedings{marquis2022,
    author="Marquis, Marie-Pier and Yaddaden, Yacine and Adda, Mehdi and Gingras, Guillaume and Corriveau-Ct{\^o}{\'e}, Michael",
    editor="Mahyuddin, Nor Muzlifah and Mat Noor, Nor Rizuan and Mat Sakim, Harsa Amylia",
    title="Automatic Honey Bee Queen Presence Detection  on Beehive Frames Using Machine Learning",
    booktitle="Proceedings of the 11th International Conference on Robotics, Vision, Signal Processing and Power Applications",
    year="2022",
    publisher="Springer Singapore",
    address="Singapore",
    pages="820--826",
    isbn="978-981-16-8129-5",
    doi="10.1007/978-981-16-8129-5_125"
}

License