News register
Search section
Search area
News Type
News type
Search date
Search word OR
List of related articles
Contents of related articles
No info was found
List of related articles
Contents of related articles
No info was found
View details
Information

11/28/2017 HYU News > Academics > 이달의연구자

Title

[Excellent R&D] Inventing Eyes for Robots

Lim Jong-woo (Professor, Department of Computer Science) winning a government project

김소연

Copy URL / Share SNS

http://www.hanyang.ac.kr/surl/L94Q

Contents
Augmented Reality, self-driving cars, and facial recognition are no longer a technology of future. Such advanced technologies are deep in our daily lives. In order for machines to properly function as they are meant to, they need something called ‘machine vision’. Machine vision (MV) is the technology and method used to provide imaging-based automatic inspection and analysis for such applications as automatic inspection, process control, and robot guidance, usually in industry. And the field that encompasses the subject is Computer Vision, which Lim majors. For December’s Researcher of the Month, News H interviewed Lim Jong-woo (Professor, Department of Computer Science) who recently won a major government project to acquire the source technology for such field. 
 
Lim is enthusiastically explaining how the technology can be applied in real lives. For example, with the structure modeling, calculating the altitude of a person's eye level (when wearing an AR/VR glasses) would be possble.


The final goal of this four-year project is to develop a high-level video situation recognition technology based on structural modeling and geometrical analysis of images acquired in extremely congested situations such as the real environment. Structural modeling of a video means to draw lines and actually structure the surrounding environment within the video, either in a two-dimensional or three-dimensional form. Up to current technology, a system can process a single object in the video or occasionally multiple objects. However, it is not yet developed for computers to recognize and analyze a ‘congested’ video with dozens of moving objects, which is often the case in real life footage. “If developed further enough, a computer would be able to track irregular paths taken by a suspect from CCTV video and alert us,” mentioned Lim. 
 
(Left) Estimation of the structure of a space through existing technology
(Right) Provisioned result of structure estimation
(Photo courtesy of Lim)


One of the ultimate goals of the project is to also integrate multi-object detection and tracking with the environment. “There are a lot of people trying to integrate detection and tracking technology,” said Lim. Because it is highly improbable for researchers to set a model human face for the computer to detect all human faces, integrating such technology with tracking a moving person is even more intricate and difficult. Nevertheless, if it does become reality, computers will be able to read the context of a specific video. For instance, because they can recognize each person, it would be able to write a storyline and understand relationships between characters in a show or a movie. 

As mentioned in the earlier part of the article, computer vision is a crucial part of augmented reality and autonomous cars. In the case of AR, the computer must be able to structure its environment to decide where to put the virtual object. Also, by such mapping, the machine can change its perspective in accordance with the user’s change of perspective. Furthermore, autonomous cars require even higher accuracy of computer vision in order to detect obstacles and prevent unwanted accidents. Unlike the facial detection of a camera app on our cellphone which is not really a matter of life and death, technology related to transportation has higher standards for that reason.
 
"I aim to research for use, rather than a reasearch for research."


Another surprising aspect of this research project plan is that the team will upload their findings on the web, free of charge as an open-source. When asked why not commercialize it, Lim answered “It is mutually beneficial for us to have the crowd test our algorithm and give feedback to us, as we cannot test it in every environment. Also, it is a trend to release algorithms open-source, because most of them fall short to be commercialized yet.” The research has begun this August and will be continued until the end of 2020. News H is looking forward to observing Lim’s progress and the social impact his team will bring. 


Kim So-yun       dash070@hanyang.ac.kr
Photos by Choi Min-ju
Copy URL / Share SNS

0 Comments