Recently, Prof. XIONG Zhiwei’s research group from School of Information Science and Technology, USTC, was conferred the Winner Award in New Trends in Image Restoration and Enhancement Workshop (NTIRE) Challenge on Spectral Reconstruction from RGB Images, CVPR 2018.

IEEE Conference on Computer Vision and Pattern Recognition (CVPR) is one of the three top international conferences in the area of computer vision. NTIRE Challenge on Spectral Reconstruction from RGB Images became an annual event since 2016, and it has shown important influence on underlying visual processing area represented by super resolution. This year, the Challenge attracted nine competition teams from three Continents including North America, Europe and Asia. Prof. XIONG’s research group proposed two spectral reconstruction programs based on 200 layers of densely converged networks and residual networks, which won the first and second place respectively and received the Winner Award.

It is worth mentioning that high-dimensional image processing based on deep neural network needs lots of computing resources, meanwhile, USTC National Engineering Laboratory for Brain-like Intelligence Technology and Applications built a GPU cluster to better support researches in artificial intelligence area. The GPU cluster now offers 400 graphics processing units for more than three hundred user clients, and it has reached nearly 200, 000 machine times each month.