Skip to content
Snippets Groups Projects
Commit cc68cc6c authored by Arfib Manon's avatar Arfib Manon
Browse files

pimpage README

parent 8e7573b8
No related branches found
No related tags found
No related merge requests found
# Detection and identification of bird calls
## Description
This git repository is made to detect and recognize birds in sound files. It is part of the project of CentraleSupélec's S6.10.06 students.
## Installation
The required Python modules are listed in the 'requirement.txt' file. Beware that CUDA necesitates a specific installation and may cause compatibility issues.
## Usage
To use the AI :
put in the input folder :
an "audio_files" folder with the audios and a csv file containing the audios you want to test. The format of the csv is :
audio_id
audio1
audio2
...
no extension on the audio_id column
Note : the csv can have multiple columns, but only the audio_id is considered
To make the training :
download and preprocess the dataset : in the terminal, put yourself in the "birdcall-detection" folder. Then type "make prepare"
Wait until the download ends. Then you can launch the training with "make train" (still in the terminal).
If you have not enough RAM, you'd better train models one by one. To do so, write (still in the terminal of the "birdcall-detection" folder) :
python train -m 1
python train -m 2
python train -m 3
python train -m 4
Note : If you want to make your own dataset, in the train/data_training folder, put an "audio_files" folder and a csv with the adequate format :
cnt,en,id,length
**,class1,audio1.mp3,**
**,class1,audio2.mp3,**
...
**,class2,audio4.wav,**
...
where audio.* is just the name of the corresponding audio file and *** are characters that needs to be there but are not important
## Acknowledgment
A huge thank you to Fred Ngole-Mboula for his help and valuable advice.
\ No newline at end of file
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment