Helmet Violation Model Prediction Helmet Violation Model Prediction is an open-source project that utilizes deep learning models to predict helmet violations based on vehicle data. The project implements several machine learning models and provides easy-to-use APIs for users to experiment with helmet violation detection.

Table of Contents Who Should Use Helmet Violation Model Prediction Why Use Helmet Violation Model Prediction Project Capabilities Requirements and Installation Training Usage Citation License Who Should Use Helmet Violation Model Prediction Beginners: Those who want to get started with deep learning for real-time object detection tasks like helmet violation detection. Engineers: Engineers interested in trying out different helmet violation models to see how deep learning can be applied to improve traffic safety. Researchers: Researchers looking for an open-source solution to easily experiment with different helmet detection algorithms and models. Why Use Helmet Violation Model Prediction Lightweight: Easy to integrate and use without heavy dependencies. Open Source: Free to use, modify, and distribute under the MIT license. Active Maintenance: The project is regularly updated with improvements, new features, and bug fixes. Simple Setup: Get started with pre-trained models or use your own datasets for training. Project Capabilities At this moment, this project includes the following models and techniques for helmet violation detection:

Type ABBRV Algorithms Description Object Detection YOLO YOLOv3 Helmet Detection Pre-trained YOLO model for helmet detection. Image Classification CNN CNN-based Helmet Detection A basic CNN model for classifying helmet violations. Transfer Learning ResNet Transfer Learning using ResNet-50 for Helmet Detection Pre-trained ResNet-50 fine-tuned for helmet detection. Hyperparameter Optimization HPO-CG Hyperparameter optimization for tuning deep learning models Optimizing model performance via HPO. For more details on each algorithm, please refer to the specific algorithm documentation provided in the repository.

Requirements and Installation System Requirements: Python >= 3.6 PyTorch >= 1.5.0 OpenCV (for image visualization and processing) Additional libraries: numpy, matplotlib, scikit-learn Installation Steps: Clone the repository:

git clone https://github.com/Hackb07/helmet_violation_model_prediction.git

Navigate to the project directory:

 cd helmet_violation_model_prediction

Install dependencies:

pip install -r requirements.txt

(Optional) If you want to use pre-trained models, download them and place them in the .latent-data/ directory.

Dataset: The project supports image and video datasets for helmet violation detection. Place your dataset in the data/ directory with subfolders for "helmet" and "no_helmet" classes, as shown below:

/data
    /train
        /helmet
        /no_helmet
    /test
        /helmet
        /no_helmet

Training To train the model, ensure that your dataset is properly structured as mentioned above. Then, run the following command to start training:

python train_model.py --data <path_to_data_directory>

You can specify the dataset directory using the --data argument.

Usage To make predictions on images or videos, use the following commands:

For Image-Based Prediction:

python predict_helmet_violation.py --image <path_to_image>

For Video-Based Prediction:

 python predict_helmet_violation.py --video <path_to_video>

Command Line Arguments: --image: Path to the image file for prediction. --video: Path to the video file for prediction. Citation If you find this project useful in your research or development, please consider citing the related papers:

bibtex Copy code @inproceedingstharun2024helmet, title = Helmet Violation Detection using Deep Learning author = Tharun Bala booktitle = Proceedings of the International Conference on AI and Machine Learning year = 2024

License The entire codebase is under the MIT license. See the LICENSE file for more details.