Back to Projects

Team Name:

GovernAI


Team Members:


Evidence of Work

SafeVision

Project Info

Team Name


GovernAI


Team Members


Megha Narchal , Aman , Niketan , Sumant

Project Description


Safevision: AI-Powered Face Mask Detection for Public Health

Safevision is an AI-driven project designed to detect and count individuals wearing face masks in real-time video feeds, assisting public health governance during crises like the COVID-19 pandemic. Using computer vision and deep learning, Safevision helps authorities monitor mask compliance in public spaces without compromising privacy.

Features

  • Real-time Mask Detection: Accurately identifies people wearing or not wearing masks in live video streams.
  • Automated Monitoring: Reduces the need for manual mask enforcement and allows efficient resource allocation.
  • Privacy-Focused: Ensures that no personal identity information is collected or stored—focused only on mask detection.
  • Actionable Data: Provides real-time statistics on mask compliance for public health authorities to make informed decisions.
  • Scalable and Customizable: Can be integrated into surveillance systems in public places like hospitals, schools, transportation hubs, and government offices.

Technologies Used

  • Deep Learning: Utilizes a fine-tuned VGG16 model for mask detection.
  • Computer Vision: OpenCV for video stream processing.
  • Python: Backend code for video stream handling and logic.

- Keras/TensorFlow: Frameworks used to train and run the deep learning model.

How Safevision Meets the GovHack2024 AI in Governance Challenge Criteria

Safevision is designed as a practical solution for public health governance, leveraging AI to detect and count individuals with and without masks in real-time video feeds. Here's how it meets the key criteria of the GovHack2024 AI in Governance Challenge:

1. Operational Efficiency

Safevision automates the detection of mask compliance in public spaces, reducing the need for manual intervention. By analyzing video feeds in real-time, it provides immediate feedback on mask usage, helping authorities quickly allocate resources to areas where intervention is needed. This makes enforcement more efficient and frees up personnel for other critical tasks.

2. Privacy and Security

Privacy is at the core of Safevision. The system does not store or process any personal identification data—its sole focus is on detecting whether individuals are wearing masks. No video footage or personal information is saved, ensuring compliance with privacy regulations like GDPR. Data is minimized and encrypted when necessary to maintain secure communication between systems.

3. Transparency and Accountability

Safevision enhances transparency by generating anonymized reports on mask compliance, which can be shared with the public. This allows communities to see how well guidelines are being followed and provides accountability in the enforcement of public health measures. The open-source nature of the project allows for scrutiny and improvement by anyone.

4. Ethical Use of AI

Ethical AI principles are embedded in Safevision’s design. The system has been trained on a diverse dataset to ensure fairness across different demographic groups, reducing bias. It provides clear and simple predictions (with or without mask), ensuring transparency in how decisions are made. Safevision promotes fairness, transparency, and accountability, aligning with ethical AI standards.

5. Building Public Trust

By ensuring transparency and protecting privacy, Safevision helps build trust between the government and the public. Citizens can access anonymized compliance data, fostering a sense of collective responsibility in public health efforts. The emphasis on privacy reassures the public that their identities are not being tracked or stored.

6. Scalability and Future Adaptations

Safevision is designed to scale effortlessly—whether deployed in small government offices or large public spaces, it handles varying loads without sacrificing performance. Its adaptable architecture allows for integration with other public health monitoring needs, such as detecting social distancing or other health measures in future emergencies. Safevision can evolve with emerging AI technologies to continue serving public interests effectively.


Conclusion

Safevision not only improves the efficiency of public health governance but does so ethically and transparently, ensuring privacy and building public trust. It's a scalable, adaptable solution for governments looking to leverage AI in real-time compliance monitoring.


Installation

To run Safevision locally, follow these steps:

  1. Clone the repository:
    bash
    git clone https://github.com/akd6203/Safevision.git
    cd Safevision

  2. Install required Python packages:
    bash
    pip install -r requirements.txt

  3. Download the trained model:
    Place the pre-trained face_mask_fine_tune_model.h5 file in the project directory. You can download the model here.

  4. Run the mask detection application:
    bash
    python mask_detection_demo.py

How It Works

  • The system uses a pre-trained VGG16 model fine-tuned for mask detection.
  • Video frames are captured using OpenCV from a webcam or video file.
  • Each frame is processed, faces are detected, and the AI model predicts whether a person is wearing a mask or not.
  • The video feed is displayed with bounding boxes around faces and mask/no-mask labels, along with real-time counters of masked and unmasked individuals.

Contributing

Contributions are welcome! If you'd like to contribute to Safevision, please fork the repository and submit a pull request.

  1. Fork the repository
  2. Create a new branch (git checkout -b master)
  3. Commit your changes (git commit -m 'Add some feature')
  4. Push to the branch (git push origin master)
  5. Open a pull request

Contact

For any inquiries or suggestions, feel free to reach out, Happy Learning!


#facemask detection #public health #ai in governance #public safety #health regulations

Data Story


Data Story: Protecting Public Health with AI

Imagine walking through a busy city square. People are rushing to catch buses, families are strolling through parks, and students are heading to their next class. Amid all this movement, something invisible lingers—the risk of spreading infectious diseases. In times of a pandemic, something as simple as wearing a mask can make a big difference in protecting not just ourselves, but those around us.

Now think about the challenge governments face in ensuring public health compliance, like mask mandates, across large populations. It's not realistic to have officials at every corner monitoring whether people are wearing masks. Manual enforcement is costly, slow, and inconsistent.

This is where AI steps in.

Our project, SafeVision, brings together the power of artificial intelligence and computer vision to help governments automatically detect mask usage in public spaces. Imagine the same bustling city square, but now equipped with AI-powered cameras that can instantly detect whether people are wearing masks. In real-time, this system counts how many people are following the guidelines, without ever compromising anyone’s privacy. No personal data is collected—just the important insight: are people protecting themselves and others by wearing masks?

Behind the scenes, our system analyzes each face it detects, determining within seconds if the person is wearing a mask or not. It's like having an extra set of eyes that works tirelessly, 24/7, to keep our public spaces safer. This data can be fed back to public health officials, helping them make informed decisions on where to focus efforts and resources.

By automating this process, SafeVision doesn’t just save time—it also builds transparency. The system could provide regular updates to the public, showing how well communities are adhering to mask mandates. This fosters trust, as citizens can see the data and understand the impact of these health measures.

We understand that AI can sometimes raise concerns about privacy and fairness. That’s why we designed SafeVision with ethical considerations in mind. The system doesn’t store personal information—it only looks for one thing: the presence or absence of a mask. And we’ve trained it on diverse datasets to make sure it works fairly for everyone, regardless of age, gender, or race.

In the end, SafeVision is more than just a tool. It’s a way to empower governments and communities to protect public health, using cutting-edge technology that respects both safety and privacy.


Evidence of Work

Video

Team DataSets

Face Mask Dataset

Description of Use Face mask detection and counting to ensure public safety.

Data Set

Challenge Entries

AI in Governance

How can governments use AI to boost efficiency and transparency in public sector operations while addressing concerns regarding ethics, data privacy, and public trust?

Go to Challenge | 35 teams have entered this challenge.

Improving the Accessibility of Online Government Services

How can we leverage AI to design inclusive digital solutions that ensure seamless and equitable access to essential services for people with disabilities?

Go to Challenge | 29 teams have entered this challenge.