BoutRightYOLO

Installation

Download anaconda https://www.anaconda.com/download

Getting Started

Follow the following steps to get started with boutRight:

Clone the repository

Clone the repository using Git, if you have a git installed in your computer. To download git go to Downloads and choose a git suitable for your operating system. You can learn more about git here. Optionally, you can install git in anaconda prompt by typing the following command:

conda install git

Once the git download is done, open a terminal or command prompt and run the following command:

git clone https://github.com/RuthKassahun/BoutRightYOLO.git

If you don't have a git installed, alternatively, you can download the zip file of this repository if you haven't installed git on your local machine.

Navigate to the Project Folder

Open Anaconda Prompt and navigate to the "boutRight" folder by using the cd command. Replace /path/to/boughtRight_yolo with the actual path to the "boutRight" folder on your local machine.

cd /path/to/boutRight_yolo

Create a Virtual Environment

Create a virtual environment using the provided environment.yml file. This YAML file specifies the dependencies and environment settings required for the project.

conda env create -f boutright_environment.yml

Activate the Environment

Activate the newly created conda environment

conda activate boutRight

Incase you want to do some sanity check if pytorch works with the configured cuda version type python once you can see these >>> type

import pytorch

torch.cuda.get_device_name(0) -- This will display the GPU device your cuda is currently accessing

Proces Dataset (Generate Spectrogram from song file)

place all your song files inside songs_for_training folder in the main boutRight_yolo folder

Prepare a virtual envrinoment for labelImg

1.1 Create a new conda envirnoment for annotating spectograms using labelImg. As LabelImg runs in python 3.8
```
conda create -n imageannotation python==3.8
```
1.2 Activate the envirnoment
```
conda activate imageannotation
```
1.3 Install LabelImg
```
pip install labelImg
```
Update your data.yaml configuration file

Open data.yaml file and change the directory path for both train and val.
```
train : {\path\to\boutRight_yolo}\dataset\images\train
val :   {\path\to\boutRight_yolo}\dataset\val
```
for example : if you saved your boutRight_yolo folder in a the following path D:\Code\boutRight_yolo, you can change it as
```
train :   D:\Code\boutRight_yolo\dataset\images\train
val :     D:\Code\boutRight_yolo\dataset\images\val
```
Open boutRight_yolo_train.ipynb jupyter notebook in visual studio code or in localhost This will go through all wav files in the given songs_for_training folder and create a new folder named images_for_training inside your main directory and add all the generated spectrograms. Run the first two cells
Open LabelImg Inside labelImg conda envirnoment type
```
labelImg
```
This will open a small window to annotate the spectrograms
Place all your annotation files inside annotations folder in the main directory, then run the third cell (Split Dataset section) from the notebook.

Train Network

Run the network - you can adjust img_size (--img), epochs (--epochs)
Once the training is done, you can run val.py to evaluate the performance of the network in your validaiton dataset.

Common known issues-

while training the network you might come across the following RunTimeError

NotImplementedError: Could not run 'torchvision::nms' with arguments from the 'CUDA' backend. This could be because the operator doesn't exist for this backend, or was omitted during the selective/custom build process (if using custom build). If you are a Facebook employee using PyTorch on mobile, please visit https://fburl.com/ptmfixes for possible resolutions. 'torchvision::nms' is only available for these backends: [CPU, Meta, QuantizedCPU, BackendSelect, Python, FuncTorchDynamicLayerBackMode, Functionalize, Named, Conjugate, Negative, ZeroTensor, ADInplaceOrView, AutogradOther, AutogradCPU, AutogradCUDA, AutogradXLA, AutogradMPS, AutogradXPU, AutogradHPU, AutogradLazy, AutogradMeta, Tracer, AutocastCPU, AutocastXPU, AutocastCUDA, FuncTorchBatched, BatchedNestedTensor, FuncTorchVmapMode, Batched, VmapMode, FuncTorchGradWrapper, PythonTLSSnapshot, FuncTorchDynamicLayerFrontMode, PreDispatch, PythonDispatcher].

Fix the issue by uninstalling torch

pip uninstall torch torchvision

then installing torchvision and torchaudio with the proper cuda verison

pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116

Model Inference

If you want to use the pretrained model and run it for your custom dataset without training, you can use boutRight_yolo_infer notebook

Open boutRight_yolo_infer.ipynb
Create a temporary folder called Temp and change the directory of temp_path to your specified directory
```
temp_path = {path/to/temp_folder}
```