Intelligent Surveillance
Overview
With the vigorous development of technologies such as AI deep learning, big data analysis, and mobile unmanned vehicles, intelligent surveillance applications have evolved from traditional fixed monitoring to unmanned vehicle patrols, and the application range has also extended from ground inspections to aerial inspections. Chunghwa Telecom Laboratories develops the multiple computer vision (CV) technologies including AI image detection and recognition, object tracking, behavior understanding and description, etc. Besides, we are committed to the research of software and hardware integration including edge computing technology (Edge AI), multiple surveillance devices (fixed cameras, mobile devices, unmanned vehicles), and the IT big data platform to create more intelligent and practical surveillance applications. It’s expected that the above research can promote surveillance applications to a new generation of full intelligent.Intelligent Surveillance
CORE TECHNOLOGY
- Edge AI License Plate Recognition and Trajectory Big Data Analysis Technology
- Traffic Law Enforcement Technology
- Unmanned Vehicle Smart Inspection Technology
Intelligent Surveillance Applications
Intelligent Surveillance
Application Status
Edge AI-based License Plate Recognition and Trajectory Analysis with Big Data Technology：We use a lightweight AI image recognition algorithm based on Edge AI technology to develop a high performance license plate recognition application with high accuracy and low cost. We combined multiple technologies such as big data analysis of vehicle information retrieval and abnormal driving trajectory and applied them to the intelligent surveillance system. The technology holds the largest share of license plate recognition in Taiwan.
Traffic Law Enforcement Technology：We use AI image/video analysis technology to build a multi-function law enforcement system in traffic scene, including vehicles failing to yield to pedestrians, running a red light, driving without following the markings, crossing prohibited lines, average speed enforcement, and illegal parking on red/yellow lines. The system combines embedded technology and deep learning algorithms to provide accurate identification of violations and applies technologies such as on-premise/multi-rental architecture on cloud and edge computing to provide county and city police with innovative law enforcement applications and 7x24 law enforcement solutions. The relevant technological achievements have been applied in New Taipei, Taoyuan, Kaohsiung, etc.. In 2022, it was awarded the Outstanding Application Award for Cloud IoT Innovation by the Cloud Computing & Association in Taiwan. In 2023, it won the Champion Award in the AI Application Contest held by Administration for Digital Industries, MODA. And in 2024, it secured a gold medal at the Invention Competition at Taiwan Innotech Expo.
Unmanned Vehicle Intelligent Inspection Technology：Utilizing advanced technologies such as unmanned vehicles (UVs), AI control, and machine vision, we have developed a comprehensive range of intelligent UV inspection services. These services cater to smart inspection applications in large facilities, energy sectors, and environmental areas. In recent years, we have actively engaged in government initiatives, including the Unmanned Aircraft System Traffic Management (UTM) project and Smart Port projects featuring 5G unmanned vehicle applications. Our focus has been on developing 5G BVLOS (Beyond Visual Line of Sight) inspection applications for both drones and surface unmanned vehicles, covering diverse inspection fields such as solar energy, transmission towers, port areas, rivers, and bridge piers. Our relevant technological achievements have garnered prestigious honors, including the 2022 Asia Communications Award for Best Smart Places, the 2023 Smart City Exhibition Innovation Application Award, 2023 Defense Application Unmanned Aerial Vehicle Challenge 2nd Place Award, and the 2024 ITS Industry Innovation Award from the Intelligent Transportation Society of Taiwan.