Adding license plate recognition function to existing HD CCTV cameras
Visionin, fire detection, machine vision, fingerprint recognition and
It has strengths in vehicle license plate recognition, etc.

At the 2018 Korea Electronics Show held at COEX in Samseong-dong, Seoul from October 24 to 27, numerous companies participated and showcased their technological prowess.

While some companies set up large booths to immediately grab the attention of visitors, there were also companies that set up small booths but captured the attention of visitors with specialized technologies. One of them was Visionin, which demonstrated how CCTV accurately captured the license plate information on a moving model train in real time.
Visionin, located in Michuhol-gu, Incheon, is a venture company established in November 2016. It was incorporated in September 2014, and in November of the same year, it received a contract to develop an environmentally adaptive image processing algorithm for intelligent vehicles. In June 2017, we developed a visual knowledge integration management interface and metadata viewer.

In October 2017, the company won the grand prize at the 2nd IoT Innovation Challenge hosted by the Ministry of Trade, Industry and Energy and sponsored by Samsung Electronics for its ‘Multi-Sensor IoT Video Fire Detector’ based on intelligent image analysis technology.

Visionin, which aims to improve the quality of human life through technological advancement, is focusing its R&D capabilities on computer vision, parallel image processing, machine learning, and embedded system technology in the era of the 4th industrial revolution. Based on this, it aims to become a global powerhouse that takes the lead in domestic production of core technologies in the fields of intelligent image surveillance, machine vision, biometrics, and smart automobiles.

At the 2018 Korea Electronics Show, where numerous companies showcased their technological prowess, Visionin demonstrated its embedded license plate recognition technology.

South Korea is a small country. It has a population of 51.82 million people living in a land area of just over 100,000㎢, and the number of registered cars is 22.53 million. As traffic congestion becomes increasingly severe, the demand for systems to crack down on illegal vehicles and vehicles occupied by criminals is increasing day by day.
Visionin's embedded vehicle license plate recognition technology, based on deep learning, detects vehicles and recognizes license plates using CNN (Convolutional Neural Network), and has embedded application cases such as NVIDIA TK1 and TX1. It can also provide vehicle license plate recognition functions to CCTVs without replacing existing cameras.

The advantages of Visionin’s embedded vehicle license plate recognition technology are as follows:

First, it has a fast recognition speed. Visionin’s embedded vehicle license plate recognition technology has secured a fast recognition speed with Visionin’s own high-speed and lightweight know-how. It is capable of recognizing vehicle license plates in FHD video at up to 15 frames per second.

And it is possible to recognize various license plates. It is possible to detect both close and long distance license plates at the same time, and it is possible to classify private, business, and special license plates. Visionin has license plate recognition technology in Korea, Europe, the United States, and other countries.

Visionin's embedded vehicle license plate recognition technology, which implements embedded deep learning using Arm's NEON technology, can be applied to crime prevention CCTV. As an edge-based solution in the intelligent video security market, it is equipped with a self-developed deep learning algorithm and supports various VMS linkage APIs.

It adopts ARM Cortex A17 1.2GHz processor, image resolution is HD (1280 x 720), processing speed is 10FPS, minimum license plate recognition size is 45 pixels, maximum recognition distance is 10m, and maximum recognition speed is 60km/h.
The embedded vehicle license plate recognition technology presented by Visionin at the 2018 Korea Electronics Show not only recognized multiple license plates simultaneously, but also had the function of sending different colored notifications depending on the vehicle type.

Visionin's Senior Researcher Jeong-hyeok Kim said, "Visionin's embedded vehicle license plate recognition technology is compact and lightweight, so it can be applied anywhere without significantly changing existing facilities," and "It also has the usability to immediately respond to new and foreign license plates in addition to license plates currently in use."

Meanwhile, Visionin also has its own technological capabilities in the fields of fire detection, machine vision, and fingerprint recognition, and is developing core technologies for various equipment in a customer-tailored manner.