We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

We present a system for accurate real-time mapping of complex and arbitrary indoor scenes in variable lighting conditions, using only a moving low-cost depth camera and commodity graphics hardware. We fuse all of the depth data streamed from a Kinect sensor into a single global implicit surface model of the observed scene in real-time. The current sensor pose is simultaneously obtained by tracking the live depth frame relative to the global model using a coarse-to-fine iterative closest point (ICP) algorithm, which uses all of the observed depth data available. We demonstrate the advantages of tracking against the growing full surface model compared with frame-to-frame tracking, obtaining tracking and mapping results in constant time within room sized scenes with limited drift and high accuracy. We also show both qualitative and quantitative results relating to various aspects of our tracking and mapping system. Modelling of natural scenes, in real-time with only commodity sensor and GPU hardware, promises an exciting step forward in augmented reality (AR), in particular, it allows dense surfaces to be reconstructed in real-time, with a level of detail and robustness beyond any solution yet presented using passive computer vision.

/pdf/kinectfusion-real-time-dense-surface-mapping-and-tracking-pl8e8pbmre.pdf

KinectFusion: Real-time dense surface mapping and tracking

Humans perceive the three-dimensional structure of the world with apparent ease. However, despite all of the recent advances in computer vision research, the dream of having a computer interpret an image at the same level as a two-year old remains elusive. Why is computer vision such a challenging problem and what is the current state of the art? Computer Vision: Algorithms and Applications explores the variety of techniques commonly used to analyze and interpret images. It also describes challenging real-world applications where vision is being successfully used, both for specialized applications such as medical imaging, and for fun, consumer-level tasks such as image editing and stitching, which students can apply to their own personal photos and videos. More than just a source of recipes, this exceptionally authoritative and comprehensive textbook/reference also takes a scientific approach to basic vision problems, formulating physical models of the imaging process before inverting them to produce descriptions of a scene. These problems are also analyzed using statistical models and solved using rigorous engineering techniques Topics and features: structured to support active curricula and project-oriented courses, with tips in the Introduction for using the book in a variety of customized courses; presents exercises at the end of each chapter with a heavy emphasis on testing algorithms and containing numerous suggestions for small mid-term projects; provides additional material and more detailed mathematical topics in the Appendices, which cover linear algebra, numerical techniques, and Bayesian estimation theory; suggests additional reading at the end of each chapter, including the latest research in each sub-field, in addition to a full Bibliography at the end of the book; supplies supplementary course material for students at the associated website, http://szeliski.org/Book/. Suitable for an upper-level undergraduate or graduate-level course in computer science or engineering, this textbook focuses on basic techniques that work under real-world conditions and encourages students to push their creative boundaries. Its design and exposition also make it eminently suitable as a unique reference to the fundamental techniques and current research literature in computer vision.

/pdf/computer-vision-algorithms-and-applications-25dn6wu83j.pdf

Computer Vision: Algorithms and Applications

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

Photo-realistic re-rendering of a human from a single image with explicit control over body pose, shape and appearance enables a wide range of applications, such as human appearance transfer, virtual try-on, motion imitation, and novel view synthesis. While significant progress has been made in this direction using learning-based image generation tools, such as GANs, existing approaches yield noticeable artefacts such as blurring of fine details, unrealistic distortions of the body parts and garments as well as severe changes of the textures. We, therefore, propose a new method for synthesising photo-realistic human images with explicit control over pose and part-based appearance, i.e., StylePoseGAN, where we extend a non-controllable generator to accept conditioning of pose and appearance separately. Our network can be trained in a fully supervised way with human images to disentangle pose, appearance and body parts, and it significantly outperforms existing single image re-rendering methods. Our disentangled representation opens up further applications such as garment transfer, motion transfer, virtual try-on, head (identity) swap and appearance interpolation. StylePoseGAN achieves state-of-the-art image generation fidelity on common perceptual metrics compared to the current best-performing methods and convinces in a comprehensive user study.

Style and Pose Control for Image Synthesis of Humans from a Single Monocular View

Differentiable rendering has received increasing interest for image-based inverse problems. It can benefit traditional optimization-based solutions to inverse problems, but also allows for self-supervision of learning-based approaches for which training data with ground truth annotation is hard to obtain. However, existing differentiable renderers either do not model visibility of the light sources from the different points in the scene, responsible for shadows in the images, or are too slow for being used to train deep architectures over thousands of iterations. To this end, we propose an accurate yet efficient approach for differentiable visibility and soft shadow computation. Our approach is based on the spherical harmonics approximations of the scene illumination and visibility, where the occluding surface is approximated with spheres. This allows for a significantly more efficient shadow computation compared to methods based on ray tracing. As our formulation is differentiable, it can be used to solve inverse problems such as texture, illumination, rigid pose, and geometric deformation recovery from images using analysis-by-synthesis optimization.

Efficient and Differentiable Shadow Computation for Inverse Problems

We present Playable Environments-a new representation for interactive video generation and manipulation in space and time. With a single image at inference time, our novel framework allows the user to move objects in 3D while generating a video by providing a sequence of desired actions. The actions are learnt in an unsupervised manner. The camera can be controlled to get the desired viewpoint. Our method builds an environment state for each frame, which can be manipulated by our proposed action mod-ule and decoded back to the image space with volumetric rendering. To support diverse appearances of objects, we extend neural radiance fields with style-based modulation. Our method trains on a collection of various monocular videos requiring only the estimated camera parameters and 2D object locations. To set a challenging benchmark, we in-troduce two large scale video datasets with significant cam-era movements. As evidenced by our experiments, playable environments enable several creative applications not at-tainable by prior video synthesis works, including playable 3D video generation, stylization and manipulation11willi-menapace.github.io/playable-environments-website.

Playable Environments: Video Manipulation in Space and Time

Digital video is ubiquitous: virtually every mobile device comes with at least one high-resolution video camera, and users often upload video to community websites -- 300 hours of video are uploaded to YouTube alone every minute. Yet, many commercial solutions for capturing, editing and browsing videos are difficult to use for end users, and hence constrict user creativity. Video capture requires framing techniques and shot planning to effectively convey the intended message and be comfortable to watch. Handheld capture with mobile devices in particular often results in shaky and wobbly footage. In addition, traditional video editing tools are not keeping pace with the proliferation of video content, and some provide little more than an image-editing interface with a timeline. Unfortunately, this rather trivial addition of the temporal axis to images does not enable users to perform complex editing tasks that change the content of videos, like adding or removing objects in a video. Furthermore, existing community video collections typically treat videos like photos by having users navigate static thumbnails, instead of visualising the spatial or temporal overlaps between videos. User-Centric Computational Videography aims to improve the quality and flexibility of capturing, editing, and exploring consumer videos. In this course, we discuss recent techniques in computer vision and graphics, and analyze how they have advanced towards this goal. By finding and exploiting inter- and intra-video content connections, these techniques make videos easier for amateur users, for example by enabling dynamic object removal in videos, and provide new empowering video experiences like content-based video browsing. We will take stock of the progress made so far on this topic, discuss current trends in the software industry as well as in research, and propose directions for future research. Our course is targeted at a broad audience: enthusiastic video users will discover cutting-edge video processing techniques that may soon find their way into consumer applications, video editors will learn about powerful approaches that break from the norm of timeline editing, and researchers will benefit from a high-level overview and analysis of user-centric video techniques. We consider the first and last quarters of our course to be most suitable for beginners, as they provide the background on existing video tools and timeline editing, and discuss user interfaces for video exploration. The middle half of our course covers more advanced video editing techniques, which are also of interest in video production.

/pdf/user-centric-computational-videography-2fi7bsqngf.pdf

User-centric computational videography

The recently introduced gravitational point set registration methods can robustly align point sets with high noise ratios. While many results with rotation and translation resolution have been presented in the literature so far, the uncertainty about the scale resolving capability of globally multiply-linked gravitational point set registration methods is remaining. To address the uncertainty, we analyse the gravitational potential energy (GPE) functional of the system with two point sets with multivariate calculus and come to the conclusion that GPE of a singularity, i.e., the state when the template collapses to a single point, always has a lower energy than the state of the optimal alignment. Moreover, the GPE as a function of scale monotonically increases and does not have equienergetic states, which we prove for several hollow and volumetric geometric primitives in R 2 and R 3 including a unit circle, a unit sphere, a unit disk and a unit ball. We perform a series of experiments with various regular shapes and different combinations of references and templates to validate our findings. The consequence is that in practice, it is highly unlikely for globally multiply-linked gravitational alignment to resolve the scale accurately. We propose ways to overcome the limitation in scale resolution, which can be considered in the next generation of gravitational approaches.

Christian Theobalt

Papers

Style and Pose Control for Image Synthesis of Humans from a Single Monocular View

Efficient and Differentiable Shadow Computation for Inverse Problems

Playable Environments: Video Manipulation in Space and Time

User-centric computational videography

Optimising for Scale in Globally Multiply-Linked Gravitational Point Set Registration Leads to Singularities