***The detection model has been completed. I am continuing to work on the mouse movements. You can follow the test codes and the process from the process directory.***
This repository contains a custom-trained YOLOv8 object detection model, built specifically for high-precision detection tasks in real-time environments.
- Trained on 5,500+ labeled images
- Includes custom datasets such as:
- Valorant characters
- Headshots
- Merged multi-label sets
- Achieved high precision and recall on real-world gameplay footage
- Optimized for real-time inference
- YOLOv8 by Ultralytics
- Python (PyTorch)
- Google Colab for training
Model was tested live using a custom-built application. Results showed significantly better performance than publicly available models, especially in identifying in-game actions and character details.
- Over 5,500 labeled images used for training
- Achieved remarkable accuracy across multiple custom classes
- Optimized for real-time inference performance
- Trained using Ultralytics YOLOv8 on a carefully curated dataset
- Classes include game-specific visuals (such as headshots, player types, etc.)
The model was trained and tested thoroughly, and outperformed several publicly available alternatives, especially in precision and recall.
Feel free to explore the repository, test the model with your own data, or use it as a starting point for your own AI projects.