AI Can Now Search for People by Description Using Surveillance Cameras

A team of Indian researchers has developed an algorithm that can independently search for people in surveillance footage based on their physical description. The system only needs information such as height, gender, and clothing color to identify individuals who match the description.

The developers say their system is similar to a search engine, but instead of searching the web like Google, it processes video streams. The search query consists of a set of physical characteristics. The scientists managed to train the algorithm to associate data about a person’s height, gender, and clothing color with what is actually happening in the video. To achieve this, they used convolutional neural network architecture, a deep learning tool that is highly effective for image recognition tasks.

In practice, this means that if you ask the system to “find men who are 180 cm tall and wearing red shirts,” the algorithm can provide a set of frames that match the request. This eliminates the need for a human to watch the entire video. Currently, the system correctly identifies 28 out of 41 people.

This is a solid result for an early-stage project, especially since the algorithm was tested on low-quality and challenging datasets, according to Next Web. The researchers promise that in the future, the list of search criteria will expand and the recognition quality will improve. In just a few days, a person being searched for could appear on footage from hundreds of cameras, with the total video material amounting to tens of thousands of hours.

About the Author

Prepared by: @iskatel007
Private Detective / Security Analyst / Information Security Specialist

• Telegram: @iskatel007
• Wickr Me: iskatel007
• Eleet: AB802089
• Threema: URT2S2X6

Telegram channel for business security information services (search and verification of any information):
https://t.me/joinchat/AAAAAEpnFMcCaUf9czghzg