Data Mining Practical Machine Learning Tools and Techniques

The use of technology has been proposed to improve decision-making in sport officials. The implementation of the video Assistant Referee (VAR) in association football is one example of how technolo...

Video assistant referees (VAR): The impact of technology on decision making in association football referees.

By transmitting texture and depth videos captured from two nearby camera viewpoints, a client can synthesize via depth-image-based rendering (DIBR) any freely chosen intermediate virtual view of the 3D scene, enhancing the user's perception of depth During wireless network transmission, burst packet losses can corrupt the transmitted texture and depth videos and degrade the synthesized view quality at the client In this paper, we propose a multiple description coding system for multi-path transmission of free-viewpoint video, with joint interview and temporal description recovery capability In particular, we encode separately the even frames of the left view and the odd frames of the right view, and transmit them as one description on one path The second description comprises the remaining frames in the two views and is transmitted over a second path If the receiver receives only one description due to burst loss in the other path, the missing frames in the other description are partially reconstructed using our frame recovery procedure First, we construct two recovery candidates for each lost pixel in a frame The first candidate is generated via temporal super-resolution from its predecessor and successor frames in the same view The second candidate is generated via DIBR from the received frame of the same time instance in the other view Next, we select the best pixel candidates one patch at a time, where an image patch corresponds to a neighborhood of pixels with similar depth values in the 3D scene Near-optimal source and channel coding rates for each description are selected using a branch-and-bound method, for given transmission bandwidth on each path Experimental results show that our system can outperform a traditional single-description/single-path transmission scheme by up to 55 dB in Peak Signal-to-Noise Ratio (PSNR) of the synthesized intermediate view at the client

Multiple Description Coding and Recovery of Free Viewpoint Video for Wireless Multi-Path Streaming

We propose to tackle the problem of RGB-D image disocclusion inpainting when synthesizing new views of a scene by changing its viewpoint. Indeed, such a process creates holes both in depth and color images. First, we propose a novel algorithm to perform depth-map disocclusion inpainting. Our intuitive approach works particularly well for recovering the lost structures of the objects and to inpaint the depth-map in a geometrically plausible manner. Then, we propose a depth-guided patch-based inpainting method to fill-in the color image. Depth information coming from the reconstructed depth-map is added to each key step of the classical patch-based algorithm from Criminisi et al. in an intuitive manner. Relevant comparisons to the state-of-the-art inpainting methods for the disocclusion inpainting of both depth and color images are provided and illustrate the effectiveness of our proposed algorithms.

/pdf/depth-guided-disocclusion-inpainting-of-synthesized-rgb-d-2zx8hprd8m.pdf

Depth-Guided Disocclusion Inpainting of Synthesized RGB-D Images

Ball trajectory data are one of the most fundamental and useful information in the evaluation of players' performance and analysis of game strategies. It is still challenging to recognize and position a high-speed and tiny ball accurately from an ordinary video. In this paper, we develop a deep learning network, called TrackNet, to track the tennis ball from broadcast videos in which the ball images are small, blurry, and sometimes with afterimage tracks or even invisible. The proposed heatmap-based deep learning network is trained to not only recognize the ball image from a single frame but also learn flying patterns from consecutive frames. The network is evaluated on the video of the men's singles final at the 2017 Summer Universiade, which is available on YouTube. The precision, recall, and $F1$ -measure reach 99.7%, 97.3%, and 98.5%, respectively. To prevent overfitting, 9 additional videos are partially labeled together with a subset from the previous dataset to implement 10-fold cross-validation, and the precision, recall, and $F_{1}$ -measure are 95.3%, 75.7%, and 84.3%, respectively. The source code and dataset are available at https://nol.cs.nctu.edu.tw:234/open-source/TrackNet/.

TrackNet: A Deep Learning Network for Tracking High-speed and Tiny Objects in Sports Applications*

Transmitting texture and depth maps from one or more reference views enables a user to freely choose virtual viewpoints from which to synthesize images for observation via depth-image-based rendering (DIBR). In each DIBR-synthesized image, however, there remain disocclusion holes with missing pixels corresponding to spatial regions occluded from view in the reference images. To complete these holes, unlike previous schemes that rely heavily (and unrealistically) on the availability of a high-quality depth map in the virtual view for inpainting of the corresponding texture map, in this paper a new Joint Texture-Depth Inpainting (JTDI) algorithm is proposed that simultaneously fill in missing texture and depth pixels. Specifically, we first use available partial depth information to compute priority terms to identify the next target pixel patch in a disocclusion hole for inpainting. Then, after identifying the best-matched texture patch in the known pixel region via template matching for texture inpainting, the variance of the corresponding depth patch is copied to the target depth patch for depth inpainting. Experimental results show that JTDI outperforms two previous inpainting schemes that either does not use available depth information during inpainting, or depends on the availability of a good depth map at the virtual view for good inpainting performance.

http://www.apsipa.org/proceedings_2013/papers/291_Joint-Texture-Reel-2942973.pdf

Joint texture-depth pixel inpainting of disocclusion holes in virtual view synthesis

This study investigates tracking a table-tennis ball rapidly from video captured using low-cost equipment for umpiring purposes. A number of highly efficient algorithms have been developed for this purpose. The proposed system was tested using sequences capture from real match scenes. The preliminary results of experiments show that accurate and rapid tracking can be achieved even under challenging conditions, including occlusion and colour merging. This work can contribute to the development of an automatic umpiring system and also has the potential to provide amateur users with open access to a detection tool for fast-moving, small, round objects.

/pdf/tracking-a-table-tennis-ball-for-umpiring-purposes-1l1gp4ehay.pdf

Tracking a table tennis ball for umpiring purposes

Incorporating spatial features with mutual information (MI) has demonstrated superior image registration performance compared with traditional MI-based methods, particularly in the presence of noise and intensity non-uniformities (INU). This paper presents a new efficient MI-based similarity measure which applies Expectation Maximisation for Principal Component Analysis (EMPCA-MI), to afford significantly lower computational complexity, while providing analogous image registration performance with other feature-based MI solutions. Experimental analysis corroborates both the improved robustness and faster runtimes of EMPCA-MI, for different test datasets containing both INU and noise artefacts.

/pdf/efficient-image-registration-using-fast-principal-component-3io07h6sns.pdf

Efficient image registration using fast principal component analysis

Table tennis is a fast sport and it is very difficult for a normal human being to manage accurate umpiring, especially in services (serves), which usually take less than a second to complete. The umpire needs to make over 30 observations and makes a judgment before or soon after the service is complete. This is a complex task and the author believes the employment of image processing and artificial intelligence (AI) technologies could aid the umpire to evaluating services more accurately. The aim of this research is to develop an intelligent system which is able to identify and track the location of the ball from live video images and evaluate the service according to the service rules. In this paper, the discussion is focused on the development of techniques for identifying a table tennis ball from match scenes. These techniques formed the basis of the ball detection system. Artificial neural networks (ANN) have been designed and applied to further the accuracy of the detection system. The system has been tested on still images taken at real match scenes and the preliminary results are very promising. Almost all the balls from the images have been correctly identified. The system has been further tested on some video images and the preliminary result is also very encouraging. It shows the system could tolerate the poorer quality of video images. This paper also discusses the idea of employing multiple cameras for improving accuracy. A multi-agent system is proposed because it is known to be able to coordinate and manage the flow of information more effectively.

/pdf/identifying-table-tennis-balls-from-real-match-scenes-using-3bzt1sq5tr.pdf

Identifying table tennis balls from real match scenes using image processing and artificial intelligence techniques

Distributed Video Coding (DVC) displays promising performance at low spatial resolutions but begins to struggle as the resolution increases. One of the limiting aspects is its 4×4 block size of Discrete Cosine Transform (DCT) which is often impractical at higher resolutions. This paper investigates the impact of exploiting larger DCT block sizes on the performance of transform domain DVC at higher spatial resolutions. In order to utilize a larger block size in DVC, appropriate quantisers have to be selected and this has been solved by means of incorporating a content-aware quantisation mechanism to generate image specific quantisation matrix for any DCT block size. Experimental results confirm that the larger 8×8 block size consistently exhibit superior RD performance for CIF resolution sequences compared to the smaller 4×4 block sizes. Significant PSNR improvement has been observed for 16×16 block size at 4CIF resolution with up to 1.78dB average PSNR gain compared to its smaller block alternatives.

/pdf/transform-domain-distributed-video-coding-using-larger-1lk4y4e2se.pdf

Patrick Wong

Papers

Joint texture-depth pixel inpainting of disocclusion holes in virtual view synthesis

Tracking a table tennis ball for umpiring purposes

Efficient image registration using fast principal component analysis

Identifying table tennis balls from real match scenes using image processing and artificial intelligence techniques

Transform domain distributed video coding using larger transform blocks