A model that is able to detect guns in images and videos.
- Build the model (YOLOv8 Ultralytics | Python)
- Integrate the model to the bakcend (Flask API)
- Build the front end application (Angular)
- Freeze requirements
Clone the project
git clone https://github.com/BecayeSoft/Guns-Detection-YOLOv8
Install the requirements
pip install requirements.txt
main.py loads the dataset from Roboflow using a DataFlow.py object,
train a YOLOv8 model using Model.py on the dataset, and exports the model.
Finally, it logs the model to Comet using Monitor.py.
A class that contains methods to download images from Google, unsplash and pixabay.
A script that download images using ImageDownLoader.py.
A class to load the dataset from Roboflow.
Additionally, it contains two methods to load a Roboflow model
trained on a specific version of the dataset, and another method to make inference.
These two were never used.
A class that contains everything to build the model.
Model.py contains methods to load or train the model, make inferences, export the model and more.
A class to monitor the model's performances using Comet.
Monitor.py contains methods to log the hyperparameters, the performance metrics and to upload the model.
By default, the model is automatically logged to Comet. This class allows us to create custom logs.
