This project demonstrates training a YOLOv8 model for object detection in a self-driving car scenario. The model is trained to detect various traffic signs and signals crucial for autonomous driving.
- Ensure you have the necessary dependencies installed. The notebook utilizes libraries such as PyTorch, Ultralytics, and Roboflow.
- for downloading ultralytics lib to get access to YOLOv8 model you can run this command in your notebook
!git clone https://github.com/Myworkss/ultralytics.git
- You may need a CUDA-enabled GPU for faster training. Verify CUDA and GPU drivers are correctly installed.
- The dataset used in this project is sourced from Roboflow, specifically tailored for self-driving car scenarios.
- The YOLOv8 model is trained using the provided dataset for 150 epochs with an image size of 640x640 pixels.
- Training metrics such as precision, recall, and mAP are displayed to assess model performance.
- The trained model is validated to evaluate its precision and detection accuracy.
- Various traffic signs and signals are evaluated individually to analyze the model's performance per class.
- The best trained model (
best.pt) is saved in theruns/detect/weightsdirectory. - Additional testing using videos and webcam inputs can be conducted using the provided model in the "Testing the Model" folder.
notebook.ipynb: Jupyter notebook containing the code for model training and validation.README.md: This file, providing an overview of the project, setup instructions, and model evaluation details.runs/: Directory containing training logs, model weights, and evaluation results.Testing the model/: This file is for testing the model we trained on videos, webcam and adjusting the detection square for final touchs.
- For further exploration and testing of the model, refer to the provided notebook and model weights.
- Additional testing scenarios such as real-time inference on videos and webcam inputs can be explored using the trained model.