News - CVTE | Dream-Future


    CVTE Research × Microsoft Research Asia: Let Machines “See” the World


    Applications like punching in, conference sign-in, face recognition, and license plate recognition make life more and more convenient. Behind these futuristic technologies is the computer vision that scientists have researched for more than ten years. How to enable computers to increase the image recognition speed and precision in the complex scenario is now the focus of the industry.

    Microsoft Research Asia (MSRA) was invited to CVTE Research for an academic exchange centering on computer vision. Microsoft scholars shared their research results and latest insights into this field. At the venue, CVTE Research and Microsoft discussed relevant topics including scientific research and application.  

    During the exchange, Dr. Peter T. Poon, Deputy Director of MSRA, made an opening speech and shared MSRA’s history and scientific research innovation results in the arena of computer vision and the latest progress in industry-university-research cooperation. He also expressed his view that the future trend for computer vision is to outperform human in the image recognition accuracy and comprehend what the computer sees.

    As one of the scholars to share at the academic exchange, Dr. Wang Jingdong, a senior researcher of the Visual Computing Group of MSRA, ACM Distinguished Member, and member of the International Association for Pattern Recognition, delivered a speech entitled “The High-efficiency Deep Neural Network Based on Interleaved Group Convolutions”.

     In terms of the geometric deformation modeling of deep neural network, Dr. Dai Jifeng, a senior researcher of the Visual Computing Group of MSRA, introduced the relevant content of “The Geometric Deformation Modeling in the Deep Neuro Network”. Then Dr. Yang Jiaolong, a researcher of the Visual Computing Group of MSRA, addressed under the title of “The Quantitative Function Learning of Deep Neural Network”.

    At the academic exchange, researchers of CVTE Research and sharing scholars conducted academic discussions on such issues as computer vision algorithm optimization. The researchers elaborated their own views over topics including interleaved group convolution algorithm and parameter quantization.

    Since it became the first signed member of the Innovation Together of MSRA, CVTE Research has conducted several sessions of academic exchange with MSRA over the subjects that interest both sides.

    In 2014, CVTE Research was founded as CVTE’s technology innovation and research institute. At the early stage of CVTE Research, the computing team was established and gathered domestic and overseas pioneering scholars for R&D innovation and application exploration. The team mainly researches face recognition and industrial three-dimensional vision.

    In the research of face recognition, the team concentrates on the implementation solutions that center on the human face, are based on the company’s major product application scenarios and include four modules -- image capture, image processing, image analysis, and system integration. It now focuses on such scenarios as the new retail, conference sign-in, and future classroom.

    In the field of industrial vision, the team has independently developed the optical detection technique, applied relevant research results to industrial production lines and carried out appearance inspection, identification, measurement on circuit components and solder sides to increase the yield.

    In the future, the team will use the visual computing accuracy as the start of its further research of lifelong algorithm learning and reinforcement learning. Meanwhile, it will further explore the understanding of visual computing on the correlation of things and quickly industrialize technological results to create value on the basis of CVTE’s industry accumulation and technological advantages in education, enterprise services, and automation industry, etc.

    This academic exchange aimed to grasp the current research trends, broaden the perspective of research and encourage the cooperation between scientific research and academia through the exchange with pioneering scholars.

    CVTE will be determined to shoulder the mission of exploring the future, focus on scientific research innovation and talent development and maintain an open and interactive academic exploration attitude to endow the CVTE Research with the immense attraction and eternal vitality for gathering elites.