A project developed using YOLOv8 for detecting cyclists and adding labels to live video. This project aims to provide a real-time detection system that enhances the safety of cyclists by utilizing advanced computer vision techniques.
This project, developed in collaboration with @serapgunes, detects cyclists in real-time and overlays labels on them. Our future plans include migrating this project to a VR (Virtual Reality) platform as Mixed Reality to enhance its perspective and adapt it to real-time travel.
The goal of our project is to provide practical solutions with high-accuracy models and use technology innovatively to improve the safety of cyclists.
- We trained the model using 8,000 labeled data augmented by the Augmentation method and YOLOv8 model.
- During the training and testing process, 70% of the data was used for training, 20% for testing, and 10% for validation.
Performance Metrics:
- Accuracy: 84.6%
- Recall: 96.5%
- Precision: 87.3%
- F1 Score: 91.7%
These results indicate that our model can detect cyclists with high accuracy and is successful in its task.
Make sure you have the following libraries installed. You can install them using the command:
pip install -r requirements.txtFor access to the trained model weights, please contact us at: ataysuatceylan@gmail.com srp.gns@outlook.com
"Stock footage used in the project introduction is provided by Videvo, downloaded from www.videvo.net."