https://github.com/alefiury/transformers-bee-species-acoustic-recognition

Official code implementation of Transformer Models Improve the Acoustic Recognition of Crop-Pollinating Bee Species

https://github.com/alefiury/transformers-bee-species-acoustic-recognition

Science Score: 36.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
  • DOI references
    Found 1 DOI reference(s) in README
  • Academic publication links
    Links to: frontiersin.org, zenodo.org
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (12.4%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Official code implementation of Transformer Models Improve the Acoustic Recognition of Crop-Pollinating Bee Species

Basic Info
  • Host: GitHub
  • Owner: alefiury
  • License: mit
  • Language: Python
  • Default Branch: main
  • Homepage:
  • Size: 930 KB
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created over 2 years ago · Last pushed over 1 year ago
Metadata Files
Readme License

README.md

Acoustic Recognition of Blueberry-Pollinating Bee Species

This repository contains the code for the paper titled Transformer Models Improve the Acoustic Recognition of Crop-Pollinating Bee Species. The study focuses on improving the recognition of bee species through their buzzing sounds using transformer-based neural network models.

Table of Contents

Introduction

The recognition of bee species through their buzzing is critical for understanding their role in pollination, particularly for crops like blueberries that rely on buzz pollination. This repository implements transformer-based models, such as Audio Spectrogram Transformer (AST), to enhance the recognition of these sounds. The project outperformed traditional CNN models, achieving state-of-the-art results in recognizing 15 different bee species based on their acoustic profiles.

Training Pipeline

Requirements

The code is implemented using Python and requires the following dependencies: - Python 3.10.14

Clone the repository and navigate to the project directory:

bash git clone https://github.com/alefiury/Transformers-Bee-Species-Acoustic-Recognition.git cd Transformers-Bee-Species-Acoustic-Recognition

Install the necessary Python libraries using the following command:

bash pip install -r requirements.txt

Dataset

The dataset used for this project comprises 518 audio samples (in which each sample is divided in smaller segments) from 15 different bee species visiting blueberry flowers in Chilean orchards. The samples contain both sonication and flight buzzing events.

Training Pipeline

The data can be found at: https://zenodo.org/records/14611302

Model Training

The repository includes the implementation of three transformer-based models:

To train the models, run the following command:

bash python3 main.py --config-path PATH_TO_CONFIG_FILE --choose-model ast

Options of models: effnet, panns, psla, ast, mae_ast, ssast

Hyperparameters

The hyperparameters for each model are configurable through the config.yaml file. Default values have been set based on the experiments conducted in the paper.

Contributors

This research was conducted by:

  • Alef Iury Siqueira Ferreira
  • Nádia Felix Felipe da Silva
  • Fernanda Neiva Mesquita
  • Thierson Couto Rosa
  • Stephen Buchmann
  • José Neiva Mesquita-Neto

License

This project is licensed under the MIT License - see the LICENSE file for details.

Owner

  • Name: Alef Iury
  • Login: alefiury
  • Kind: user
  • Location: Goiânia, Goiás - Brasil
  • Company: Universidade Federal de Goiás

Machine learning researcher and computer science undergraduate student focusing mostly on automatic speech recognition, sound event detection and bioacoustics.

GitHub Events

Total
  • Push event: 4
Last Year
  • Push event: 4