Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

From the Publisher: 
The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

Color image segmentation: advances and prospects

When considering person re-identification (re-ID) as a retrieval process, re-ranking is a critical step to improve its accuracy. Yet in the re-ID community, limited effort has been devoted to re-ranking, especially those fully automatic, unsupervised solutions. In this paper, we propose a k-reciprocal encoding method to re-rank the re-ID results. Our hypothesis is that if a gallery image is similar to the probe in the k-reciprocal nearest neighbors, it is more likely to be a true match. Specifically, given an image, a k-reciprocal feature is calculated by encoding its k-reciprocal nearest neighbors into a single vector, which is used for re-ranking under the Jaccard distance. The final distance is computed as the combination of the original distance and the Jaccard distance. Our re-ranking method does not require any human interaction or any labeled data, so it is applicable to large-scale datasets. Experiments on the large-scale Market-1501, CUHK03, MARS, and PRW datasets confirm the effectiveness of our method.

/pdf/re-ranking-person-re-identification-with-k-reciprocal-3erzmilgv1.pdf

Re-ranking Person Re-identification with k-Reciprocal Encoding

It is useful for automatic video shooting in a lecture room to estimate the location of a speaker in the lecture room. The captured videos are used for distance learning and lecture archiving systems. In order to estimate the location of a speaker in a wide lecture room, multiple cameras and multiple microphones are used. However, it is difficult to estimate the precise location of a speaker using only visual or acoustic sensors because of calibration problems, noise, and other interference. Therefore, we propose a method that integrates audio and visual information from a speaker in the lecture room. A lecturer’s cell and a student’s cell ared introduced as a unit of estimation of the location of a speaker. We defined 120 cells in a real lecture room and our multi-modal method were applied to the cells. The estimation accuracy of the location of a speaker is sufficient for automatic video shooting of a speaker in a lecture room by our integrating method.

Audio and Visual Information Integration for Speaker’s Localization in Automatic Shooting of Lecture

Nowadays, many photos are posted on social photo sharing web sites (e.g. Flickr, Instagram). The number of the shared geo tag photos in a certain attraction does not correlate to the number of visitors because the photographic activity is influenced by multiple factors. The estimation of the number of tourists in a certain attraction from the geo tagged photos is useful for marketing and tourism policy making. In this paper, from the viewpoint of tourism information, we propose a photographic activity model to estimate the number of tourists in the attraction and examine the model by using the regression analysis. We analyze the dataset consists of GPS trajectory and the photographic history of foreign tourists who traveled in Kyoto area in Japan. As the results, we show that the proposed photographic activity model indicates high goodness of fit and correlation to the dataset.

Regression Analysis of Tourist Photographic Activity Using Geo-tagged Photos andGPS Trajectory

We propose a new method to acquire accurate 3D shape of human body . Using the light stripe tri- angulation method, we can acquire accurate 3D shape of a motionless subject in a dozen of seconds. When the subject moves during the scanning, the acquired shape would be distorted. Usually, human move slightly to keep the balance, such motion is called trunk sway. Our method corrects the distorted shape with measured trunk sway during the scanning. Experimental results show the accuracy of our measurements. Trunk sway degrades the accu- racy of light stripe triangulation from 2mm to 10mm. We can maintain the accuracy as good as 2mm by applying our method.

https://www.iiyama-lab.org/publication/miru2004-funatomi.pdf

Accurate 3D scanning of trunk swaying human body

Blind deconvolution (BD) is the problem of restoring sharp images from blurry images when convolution kernels are unknown. While it has a wide range of applications and has been extensively studied, traditional shift-invariant (SI) BD focuses on uniform blur caused by kernels that do not spatially vary. However, real blur caused by factors such as motion and defocus is often nonuniform and thus beyond the ability of SI BD. Although specialized methods exist for nonuniform blur, they can only handle specific blur types. Consequently, the applicability of BD for general blur remains limited. This paper proposes a shift-variant (SV) BD method that models nonuniform blur using a field of kernels that assigns a local kernel to each pixel, thereby allowing pixelwise variation. This concept is realized as a Bayesian model that involves SV convolution with the field of kernels and smoothing of the field for regularization. A variationalBayesian inference algorithm is derived to jointly estimate a sharp latent image and a field of kernels from a blurry observed image. Owing to the flexibility of the field-of-kernels model, the proposed method can deal with a wider range of blur than previous approaches. Experiments using images with nonuniform blur demonstrate the effectiveness of the proposed SV BD method in comparison with previous SI and SV approaches. key words: blind deconvolution, deblurring, shift-variant, variational Bayes

Shift-Variant Blind Deconvolution Using a Field of Kernels

In solving pattern recognition problem in the Euclidean space, prototypes representing classes are de ned. On the other hand in the metric space, Nearest Neighbor method and K-Nearest Neighbor method are frequently used without de ning any prototypes. In this paper, we propose a new pattern recognition method for the metric space that can use prototypes which are the centroid of any three patterns in a class. This method is based on the theorem that four points of any metric space can be embedded into the Euclidean space by an appropriate metric transform.

Michihiko Minoh

Papers

Audio and Visual Information Integration for Speaker’s Localization in Automatic Shooting of Lecture

Regression Analysis of Tourist Photographic Activity Using Geo-tagged Photos andGPS Trajectory

Accurate 3D scanning of trunk swaying human body

Shift-Variant Blind Deconvolution Using a Field of Kernels

Pattern Recognition Method for Metric Space by Four Points Embedding.