Our project tackles the issue of traffic congestion, especially during significant events such as festivals and sports competitions. With the rapid expansion of cities in Saudi Arabia and the increasing frequency of large gatherings, efficient traffic management is crucial. We plan to implement computer vision techniques to identify and forecast different traffic conditions, allowing for real-time modifications to traffic systems. The proposed approach employs deep learning models and technuiques, Specifically Computer vision and LLM to monitor and improve traffic flow, ultimately aiming to alleviate congestion and enhance the overall traffic experience during major events.
The data used in this project were gathered manually from platforms and social media across the internet. The data reched to 1200 images seperated into 3 categories:
- Caused traffic (400 img)
- Regular traffic (400 img)
- No traffic (400 img)
Our project uses computer vision and natural language processing. It has three main steps: training a model with images, analyzing videos in real-time, and integrating an AI chatbot.
As mentioned before, our dataset included over 1,200 images that were manually sorted into three categories: "Caused traffic," "Regular traffic," and "No traffic" These images were used to train a convolutional neural network (CNN) with Keras. Once a baseline accuracy was achieved, video data was added for real-time analysis. A custom Flask application processed the videos frame-by-frame with a finetuned YOLO (You Only Look Once) model. The results were displayed with OpenCV annotations to show different types of congestion. Lastly, a LangChain-based chatbot, using the llama3-8b-8192 language model, was added to give contextaware answers to user questions about traffic patterns. These methods creates a strong system for accurately monitoring and responding to traffic conditions that will help to maintain during major events.
- Clone the repository:
git clone https://github.com/shhouuq/Baseer-Computer-vision-and-LLM-.git
Use the package manager pip to install requirements.
pip install -r requirements.txtclick here to load the classification model.
If you would like to contribute to this project, please fork the repository and submit a pull request. Make sure to update the tests as appropriate.