Monthly Archives: November 2016

MCL members, Yuzhuo Ren and Wenchao Zheng presented their papers at Asian Conference on Computer Vision (ACCV) 2016.

Yuzhuo’s presented paper is entitled “a coarse-to-fine indoor layout estimation (CFILE) method”. Here is an abstract of the paper:

Existing solutions to indoor layout estimation problem largely rely on hand-craft features and vanishing lines, and they often fail in highly cluttered indoor rooms. The proposed coarse-to-fine indoor layout estimation (CFILE) method consists of two stages: 1) coarse layout estimation; and 2) fine layout localization. In the first stage, we adopt a fully convolutional neural network (FCN) to obtain a coarse-scale room layout estimate that is close to the ground truth globally. The proposed FCN considers combines the layout contour property and the surface property so as to provide a robust estimate in the presence of cluttered objects. In the second stage, we formulate an optimization framework that enforces several constraints such as layout contour straightness, surface smoothness and geometric constraints for layout detail refinement. Our proposed system offers the state-of-the-art performance on two commonly used benchmark datasets.

Wenchao’s paper is entitled “Object Boundary Guided Semantic Segmentation”. Here is a brief summary of the paper:

Recent development in fully-convolutional neural network (FCN) has enabled accurate pixel-level in semantic segmentation. However, segmentation details are lost since the object boundary information is not exploited. Our proposed object boundary guided FCN (OBG-FCN) is able to integrate the distinct properties of object shape and class features elegantly in a fully convolutional way with a designed masking architecture. We show that the end-to-end trainable OBG-FCN system offers great improvement in optimizing the target semantic segmentation quality.

Congratulations to Yuzhuo and Wenchao for their quality research work!

The MCL joined a collaborative project to build a large-scale subjective video quality database. The database was proposed to boost a major breakthrough in video coding and processing. On one hand, revolutionary ideas rather than fine-tuning patches are highly expected to accommodate increasing video traffic. On the other hand, PSNR has been the dominant distortion metric for many years, but it has also been criticized for not correlating well with perceptual quality. With this database, perceptual coding is promising to lead to numerous R&D opportunities and revolutionary research with machine learning tools.

The database consists of 200 raw sequences with a duration of 5 seconds, encoded by H.264/AVC with fixed QP as bit rate control method. They are available in 5 resolutions from 3840×2160 to 540×360. Around 1000 students participated in the subjective test and it took around 7000 hours to get sufficient samples about 3 JND points. The database will be freely available for downloading for scientific purposes.

The projected was supported by 4 major multimedia companies, Netflix, Huawei, MediaTek, and Samsung. Meanwhile, 6 universities at Shenzhen joined the project, Shenzhen Institutes of Advanced Technology (Chinese Academy of Science), Shenzhen University, Graduate School at Shenzhen (Tsinghua University), Peking University Shenzhen Graduate School, The Chinese University of Hong Kong (Shenzhen), City University of Hong Kong.

We would like to give special thanks to the participating companies, institutes and universities.

Convolutional neural networks (CNNs) have received a lot of attention in recent years due to their superior performance in computer vision benchmarking datasets. Yet, little theory has been developed to explain the underlying principle of CNNs.

Professor C.-C. Jay Kuo, Director of the Media Communication Lab, has recently published important theoretical results about two CNN fundamental properties. They are: 1) why a non-linear activation function is essential at the filter output of every intermediate layer? and 2) what is the advantage of the two-layer cascade system over the one-layer system? To answer these two questions, he developed a mathematical model called the “REctified-COrrelations on a Sphere” (RECOS).

Professor Kuo said, “CNNs need to store the knowledge learned from a large amount training data somewhere. The only places to store the knowledge are the converged filter weights. They must play a critical role in CNN understanding.” Professor Kuo coined a new term “anchor vectors” for the converged filter weights since they serve as a set of vectors for an arbitrary input vector to project onto at one layer. The projected values are the response from the previous layer while their rectified values serve as the input to the next layer. In his paper entitled with “Understanding Convolutional Neural Networks with A Mathematical Model”, Professor Kuo used the anchor vector concept to explain the necessity of nonlinear activation, analyze the behavior of a two-layer RECOS system, and compare it with its one-layer counterpart. He used the LeNet-5 applied to the MNIST dataset as an illustration example throughout the paper.

Professor Kuo emphasized that the new “anchor vector” viewpoint will lead to more interesting research in the near future. His paper has recently been accepted for publication [...]