The application has been deployed in streamlit cloud, try it for yourself!!
Link -- https://ai-traffic-eye.streamlit.app/
I utilized the Selenium library in Python for automating web scraping. The scraping process posed challenges due to frequent Chrome updates that hindered traditional methods. The most demanding aspect involved configuring the chromedriver to navigate through thumbnails and identify the image source to prevent downloading of the thumbnails.
Around 307 images were carefully selected and manually inspected. To ensure consistent input dimensions for the YOLO model, resizing was performed on all images. Roboflow was utilized to manually draw bounding boxes. Additionally, image augmentation techniques were employed to expand the training dataset, enhancing the model's learning capabilities.
The Ultralytics YoloV8 nano model custom trained on dataset via Roboflow API dairectly on T4 GPU in Google Colab.