The 13th Asian Conference on Computer Vision (ACCV 2016) was held from Nov 20th to Nov 24th at Taipei International Convention Center (TICC), which is one of the biggest international conference on the “general AI” field. “Computer Vision” is a technique in general AI to study how to help computers “understand” what they “see”. Its applications include video content analysis, face detection and recognition, and self-driving cars etc.  

After 16 years, ACCV was held in Taipei again, and was supported by academics at home and abroad. Professor Yang from University of California, Professor Zha from Peking University, and Professor Hong from National Taiwan University were the chairmen of ACCV 2016. Professor Lai, Professor Sun from National Tsing Hua University, Professor Chuang, Professor Chen from National Chiao Tung University, Professor Chen, Researcher Wang, and Researcher Lin from Academia Sinica, and Professor Hsu from National Taiwan University of Science and Technology all participated in the organizing team, and hopes that the conference can facilitate exchange of thoughts in computer vision technology and enhance the level of this technology.   The computer vision and AI industry in Taiwan is still at the embryonic stage. The sponsors of ACCV 2016 include video searching company Viscovery, industrial detection company Utechzone, intelligent monitoring company UmboCV, and intelligent Ad company Appier; they all had booth stands at the conference with demos and interactions for the 500 participants at home and abroad. Furthermore, Viscovery sponsored and held the Movie Night event for ACCV participants to experience the movie with the most advanced 3D shooting technique, which can only be played in 5 cinemas in the world——Billy Lynn’s Long Halftime Walk. The technology that the director Ang Lee used is also related to issues mentioned at ACCV, bringing up discussions and interactions among the participants.


The academic research in ACCV 2016 is closely related to the latest technology and challenges faced in the industry. For example, the popular topic, self-driving cars, is what Professor Sun from NTHU has been studying; his research can help the car to predict the possibility of other car crashes to prevent itself from being hit. Professor Yang from University of California studied how to analyze video scenes and classify them with fewer manual tags. Researcher Liu from Academia Sinica makes convolution calculation from deep learning more flexible so that it can be applied on more solutions. Professor Chen from NTHU can accrately separate objects in images with few information. Moreover, researcher from Taiwan computer vision industry also shared his research during the conference: Dr. Chiu from Viscovery showed his research while he was studying in Germany and proposed a consumer stereo video segmentation benchmark as well as an ensemble method. This task of stereo video segmentation can be the basis for further applications, e.g. action recognition and surveillance scenarios.


Taiwan is still in the beginning stage of the revolution of computer vision, machine learning, and deep learning. ACCV 2016 not only promoted the development of Taiwan’s computer vision research, but also became a reference of Taiwan’s industrial transformation and industry-academic cooperation in the future.