| I&T Solution |
Intrusion Detection
(REF: S-0709) |
| Trial Project |
|
| Solution Feature |
- The Lidar camera is used to capture and recognize the depth map of a person. It can be based on a dynamic tracking algorithm and a depth map recognition algorithm to detect whether the object is a human or an animal, and protect privacy.
- Lidar cameras are deployed along the light rail to detect people that enter the dangerous area. If someone breaks into the pre-drawn dangerous area, the system will automatically alarm, and transmit the video pictures and screenshots of the intruders to the administrator.
- We can deploy a private server locally to analyze real-time video streams, or deploy a cloud server to transmit the depth map captured by the camera through the 5G network. It can be decided according to the actual situation of the project.
- Our algorithm can also identify and alert intruded objects in bad weather such as fog or rain
- When a person or a non-background object appears in the dangerous area, our system shall identify and analyze it within 10 seconds, and then alarm.
|
| Trial Application and Expected Outcome |
- The IP in 3D point cloud processing, artificial intelligence algorithms, combined with self-developed Lidar products, focus on privacy protection (non-RGB image acquisition), large angles and mask conditions analysis of the face and human behavior in the video. This series of IP is used in this project.
- We will deploy lidar cameras along the light rail platforms and conduct intrusion tests for people entering dangerous areas. The desired result is that an alarm and related video screenshots shall be given within 10 seconds. The test will help us further shorten the alarm time.
- Allowing testers to enter and exit dangerous areas with high frequency, and test whether our system can alarm in time when facing a large number of frequent intruders. The test can help us optimize the detection performance when multiple people enter the alarm area at the same time.
- Test people of different body types and builds, including different people at different age, the elderly, children, women, young men, etc. The algorithm is trained based on the test results of different people, so that it can quickly identify various groups of people and alarm, as well as recognize common animals or moving objects.
- Test the system in extreme weather conditions such as fog or rain to detect the degree of influence of different weather on the recognition accuracy, and collect data to train and optimize our algorithm.
|
| Info on I&T Solution Provider |
|
