By NUL Research and Innovations
A student from the National University of Lesotho (NUL), Mr. Thabang Nkoko has developed a system that identifies criminals or wanted people using their facial features.
Nkoko says that the high rate of criminal activities that happen in Lesotho bother him so much that with every newspaper or social media post from the police reporting on crime, he would always wonder how he can assist in reducing the numbers of people who essentially “get away with crime” or it takes a very long time for the police to find and arrest.
This is a student that has always had a passion for electronics.
“How can I use my skills to provide a tangible solution that will not only improve our police services but also help the community?” Mr. Nkoko thought to himself.
He then decided to develop a facial recognition system. This system will allow the police to identify “wanted people” from a large crowd. This facial recognition happens in real-time and uses OpenCV and face recognition libraries to capture and process video streams. Each frame of the video is analyzed to identify and encode facial features which will then be compared to the preloaded files which contain the criminals faces.
You may be wondering “how exactly does this system work?”, so let’s paint this picture for your understanding and better appreciation of the system: This system has been installed and camera’s linked to the system are placed at one of the malls in Maseru. A wanted criminal enters the building and is heading to a clothing store which is also where one of the camera has been mounted. The camera recognizes the face of the criminal out of the 53 people that passed that spot at that time. The system sends a notification to the police via telegram with the criminal’s name, location and how accurate the match may be.
The database is carefully compiled and organized so that it can easily serve as the foundation for differentiating between ordinary citizens and wanted individuals.
The system features : 1. Real-Time Face Detection and Recognition: The system processes live video feeds, detecting faces in real-time and cross-referencing them with a database of known criminals.
2. Pre-Trained Models for High Accuracy: The face recognition library uses pre-trained models to get facial features. These models have been trained on advanced datasets, which enables the system to achieve high accuracy even in challenging conditions, such as poor lighting or crowded scenes.
3. Instant Notifications via Telegram: After detecting a match, the system takes a snapshot of the suspect, it makes notes which include: the criminal’s name, and calculates the confidence level of the match i.e. how accurate the match is. A notification, complete with the image, timestamp, and location data, is then sent to a designated Telegram chat, which will ensure that law enforcement receives real-time alerts. This feature is important for rapid response, potentially preventing crimes before they escalate.
4. Location Awareness: The system integrates geolocation functionality using the Geocoder library, capable of retrieving real-time location data. In scenarios where location data is unavailable, it defaults to predefined coordinates, ensuring that every alert contains actionable information.
5. Scalable and Cost-Effective Deployment: By using the power of modern micro-controllers and single-board computers, such as the Raspberry Pi, the system can be installed across multiple locations without incurring high costs or complexity. This is essential for wide-scale implementation across Lesotho, particularly in high-risk areas like malls, banks, and border gates.
6. Future Potential and Integration: Looking ahead, the system’s modular architecture allows for future improvements, including integration with drones for aerial surveillance and GPS-based tracking for broader surveillance capabilities.
“With the support of the government or other interested entities, this system can be deployed on a national scale, providing law enforcement with a powerful tool to combat crime and enhance public safety.” Nkoko said in closing.
For more information, please contact Mr. Nkoko on +26659071616.
