LILIN seizes AI trend in security and IoT to deliver cutting-edge solutions

From the experience of promoting AI development, the Industrial Development Bureau (IDB) under Taiwan’s Ministry of Economic Affairs addresses the shortage of manpower, the pain points of industrial development, and the needs for industrial transformation by introducing a dual development model that combines top-down (theme-based) and bottom-up (needs-based) approaches.

The model links industry associations and identifies companies with AI technology to explore fields and industries in need of AI applications. Through guidance and subsidy programs, the project promotes empirical cases of AI. Through coaching mechanism, large companies are teamed up with startups. Large companies propose their R&D needs, and startups can have the opportunity to execute the concepts and become a part of the supply chain. This also helps large companies reduce R&D costs.

Through such mechanism, Taiwan has successfully cultivated many outstanding AI companies. AI use is rising in Asia. In particular, inquiries and deployment are especially obvious in one of the region’s strongest verticals – transportation. One company that has the technology and use cases in the region is LILIN. The company will discuss their solutions and projects in an upcoming webinar organized by the Industrial Development Bureau (IDB) and the Institute for Information Industry.

Aida – LILIN’s AI technology
 

AI has evolved from a mere concept to reality. LILIN has seized the AI trend in security and IoT to develop cutting-edge solutions that meet different vertical needs. The company’s AI, Aida, has various applications which are discussed as follows.
 

Traffic management

 
Aida can be especially beneficial in traffic management. As congestion and traffic violations become more rampant, Aida can play a key role in solving these issues. In particular, Aida can work wonders in the following areas.
 
License plate recognition: Aida can recognize plates with up to 99.6 percent accuracy, even when vehicles are traveling at high speeds in multiple lanes. Aida also supports multinational license plate recognition and multi-color license plate recognition.
 

Traffic flow management: When congestion begins to form, Aida can detect it and send alerts to traffic authorities. Aida can also detect multiple objects, including pedestrians, bicycles, cars, motorbikes, buses, trucks, trailers and pick-up trucks, to help with related management and planning.
 
Law enforcement: Aida can detect traffic violations, such as running red lights, making illegal turns or illegal entry into highways by wrong vehicles. It can also detect vehicles illegally parked at off-limits areas to help with fine issuance afterwards.
 
“Traffic law enforcement is an interesting topic for LILIN’s customers. It is quite difficult for linking behavior detection camera and number plate recognition camera at the same time. Once the violation behavior is detected, programmatically activating the camera to capture the number plate becomes a challenge at high speed for both day and night. LILIN’s traffic enforcement solution provides full automation for ticket generation,” said Steve Hu, Product Manager at LILIN.
 

Post-pandemic applications

 
LILIN’s Aida also has post-pandemic applications for end users in different verticals. During and after COVID-19, end users are faced with certain disease control requirements. As such, Aida can help using the following features.
 
Mask detection: Aida’s mask detection function can be beneficial for user entities where wearing a mask is mandatory. The AI can detect if someone is wearing a mask, not wearing one or half-wearing it. It can be run on but an I3 CPU-supported PC; the user does not have to acquire fancy hardware.
 
Social Distancing: Aida can be used to calculate people density and support global counters for counting people in a zone. An alarm will be sent if a zone exceeds its density threshold. This can help ensure social distancing, which is a main disease prevention measure.
 

Examples from the field

 
LILIN’s AI is not for show; it’s been deployed in real projects across the globe. Known for its strong end-to-end solution capabilities, LILIN is able to combine its AI with IP cameras, edge servers and backend systems to deliver total solutions that help meet different vertical needs. Aida’s traffic management application, for example, can be found in different projects in the world, including actual deployment in Kaohsiung, pre-installation in Italy and as proof-of-concept in California.
 
According to Hu, the APAC market especially exhibits strong potential for the AI traffic solution. “LILIN has focused on APAC and Vietnam markets for decades. After LILIN provided license plate recognition and traffic management systems, these markets have greater demands for AI traffic law enforcement solution including automated ticketing system for violations such as running red lights, lane crossing, left-and-right turns and illegal parking. The ticketing system was time-consuming for government operators. The solution is able to help traffic management of cities in APAC,” Hu said.
 
One case study worth mentioning is an ALPR project in Ho Chi Minh City, which needs a solution that can read plates quickly as drivers often drive at high speeds, and motorbikes don’t always follow road codes.
 
“Ho Chi Minh is a big city so for the first stage, eight 4-channel servers were installed for traffic monitoring and management. ALPR uses a lot of resource so we have to optimize the available hardware for the LPR software. There were a lot of parameters to set up. With our local engineering team onsite to provide support and fine-tuning, we were able to improve accuracy from 75 to 95 percent within one month. 95 percent is a very good number when it comes to ALPR,” explained Aloun Kongmany, Project Manager at LILIN, in an earlier interview with asmag.com.
 

A future that lies in edge computing

 
As for the future of AI, LILIN believes it lies in edge computing where on-premises AI cameras and edge servers can process the bulk of the video data. This reduces data transmission bandwidth and allows quicker response by the end user.
 
“Distributed cameras with powerful GPU embedded SoCs can perform AI recognition. The AI recognition results can be analyzed by a remote thin cloud or thin VMS. It is also an environmental friendly system design,” Hu said. “LILIN starts to release 7 series of AI cameras, including AI PTZ, to the industry.”

LILIN’s official name is Merit LILIN. And indeed, it has demonstrated the merit of AI in different projects, showing how its Aida can help end users achieve better security and operational efficiency. With AI being an important topic in APAC, it will be one of the focus in our upcoming webinar, “Result-focused AI: Examples from the Field,” on June 30 at 3 p.m. where LILIN’s Hu will present on the subject, “Near-edge and edge AI for traffic enforcement.” To register, click here.
                                                                 
*Industrial Development Bureau, MOEA advertisement ​經濟部工業局廣告