Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/atharv-naik/whale-call-classification
A CNN model for classifying whale calls
https://github.com/atharv-naik/whale-call-classification
audio-feature-extraction audio-processing convolutional-neural-networks deep-learning ffmpeg tensorflow
Last synced: 21 days ago
JSON representation
A CNN model for classifying whale calls
- Host: GitHub
- URL: https://github.com/atharv-naik/whale-call-classification
- Owner: atharv-naik
- License: mit
- Created: 2023-02-10T16:54:18.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2024-10-01T16:37:08.000Z (4 months ago)
- Last Synced: 2024-11-06T17:54:26.180Z (2 months ago)
- Topics: audio-feature-extraction, audio-processing, convolutional-neural-networks, deep-learning, ffmpeg, tensorflow
- Language: Jupyter Notebook
- Homepage:
- Size: 2.61 MB
- Stars: 3
- Watchers: 1
- Forks: 0
- Open Issues: 4
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Whale Call Classification
This project aims to classify Blue whale calls into A-calls or not based on a training dataset provided as a part of ISI DataFest Integration 2023. A-Calls are the most commonly heard vocalization of Blue whales and play an important role in their communication and behavior.
The project uses a Convolutional Neural Network (CNN) to classify the Blue whale calls. The CNN model is trained on a training dataset consisting of ~2600 samples and validated on a ~10% samples. The trained model was then tested on a separate testing dataset consisting of ~1000 samples.
## Installation
Clone the repo and cd into it. To install the required packages, run the following command:
```
pip install -r requirements.txt
```
Download the dataset from the competetion link [here](https://www.kaggle.com/competitions/datafestintegration2023/data) or using the kaggle API, run:
```
kaggle competitions download -c datafestintegration2023
```
Make sure the dataset is in the same directory as the working directory.
## Usage
First preprocess the data. Install ffmpeg which is a dependancy for ```make_aiff.py``` and then preprocess the data. Run the following commands:
```
sudo apt install ffmpeg
python make_aiff.py
python preprocess.py
```
To train the model, run the following command:
```
python train.py
```
## Dataset
The dataset consists of labeled audio files of Blue whale calls in WAV format. The files are categorized as A-call or not based on their spectrogram.
## Model Architecture
The CNN model consists of 3 convolutional layers, followed by 2 fully connected layers, and an output layer. The model is trained with binary cross-entropy loss and Adam optimizer.
## Results
The model achieved an accuracy of 97% on the test set.
## Authors
- [@atharv-naik](https://www.github.com/atharv-naik)
- [@Arg-10](https://www.github.com/Arg-10)
- [@SomnathRoy123](https://www.github.com/SomnathRoy123)
## License
This project is licensed under the [MIT](LICENSE) license.
[](https://github.com/Arg-10/Whale-Call-Classification/blob/main/LICENSE)