Multilayer neural networks trained with the back-propagation algorithm constitute the best example of a successful gradient based learning technique. Given an appropriate network architecture, gradient-based learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns, such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional neural networks, which are specifically designed to deal with the variability of 2D shapes, are shown to outperform all other techniques. Real-life document recognition systems are composed of multiple modules including field extraction, segmentation recognition, and language modeling. A new learning paradigm, called graph transformer networks (GTN), allows such multimodule systems to be trained globally using gradient-based methods so as to minimize an overall performance measure. Two systems for online handwriting recognition are described. Experiments demonstrate the advantage of global training, and the flexibility of graph transformer networks. A graph transformer network for reading a bank cheque is also described. It uses convolutional neural network character recognizers combined with global training techniques to provide record accuracy on business and personal cheques. It is deployed commercially and reads several million cheques per day.

Gradient-based learning applied to document recognition

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

There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality.

Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

I and i

We study the question of feature sets for robust visual object recognition; adopting linear SVM based human detection as a test case. After reviewing existing edge and gradient based descriptors, we show experimentally that grids of histograms of oriented gradient (HOG) descriptors significantly outperform existing feature sets for human detection. We study the influence of each stage of the computation on performance, concluding that fine-scale gradients, fine orientation binning, relatively coarse spatial binning, and high-quality local contrast normalization in overlapping descriptor blocks are all important for good results. The new approach gives near-perfect separation on the original MIT pedestrian database, so we introduce a more challenging dataset containing over 1800 annotated human images with a large range of pose variations and backgrounds.

/pdf/histograms-of-oriented-gradients-for-human-detection-3nteinziuq.pdf

Histograms of oriented gradients for human detection

Object detection performance, as measured on the canonical PASCAL VOC dataset, has plateaued in the last few years. The best-performing methods are complex ensemble systems that typically combine multiple low-level image features with high-level context. In this paper, we propose a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 30% relative to the previous best result on VOC 2012 -- achieving a mAP of 53.3%. Our approach combines two key insights: (1) one can apply high-capacity convolutional neural networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data is scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, yields a significant performance boost. Since we combine region proposals with CNNs, we call our method R-CNN: Regions with CNN features. We also present experiments that provide insight into what the network learns, revealing a rich hierarchy of image features. Source code for the complete system is available at http://www.cs.berkeley.edu/~rbg/rcnn.

/pdf/rich-feature-hierarchies-for-accurate-object-detection-and-22fvjlggyg.pdf

Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation

http://www.ri.cmu.edu/pub_files/2011/1/Biomaterials_Journal_Jan11.pdf

Spatially directed guidance of stem cell population migration by immobilized patterns of growth factors

A mobile robot is a vehicle that navigates autonomously through an unknown or partially known environment. Research in the field of mobile robots has received considerable attention in the past decade due to its wide range of potential applications, from surveillance to planetary exploration, and the research opportunities it provides, including virtually the whole spectrum of robotics research from vehicle control to symbolic planning (see for example [Har88b] for an analysis of the research issues in mobile robots). In this paper we present our investigation of some the issues in one of the components of mobile robots: perception. The role of perception in mobile robots is to transform data from sensors into representations that can be used by the decision-making components of the system. The simplest example is the detection of potentially dangerous regions in the environment (i. e. obstacles) that can be used by a path planner whose role is to generate safe trajectories for the vehicle. An example of a more complex situation is a mission that requires the recognition of specific landmarks, in which case the perception components must produce complex descriptions of the sensed environment and relate them to stored models of the landmarks.

/pdf/3-d-vision-techniques-for-autonomous-vehicles-4eb182vto8.pdf

3-D vision techniques for autonomous vehicles

We present a multi-layered display that uses water drops as voxels. Water drops refract most incident light, making them excellent wide-angle lenses. Each 2D layer of our display can exhibit arbitrary visual content, creating a layered-depth (2.5D) display. Our system consists of a single projector-camera system and a set of linear drop generator manifolds that are tightly synchronized and controlled using a computer. Following the principles of fluid mechanics, we are able to accurately generate and control drops so that, at any time instant, no two drops occupy the same projector pixel's line-of-sight. This drop control is combined with an algorithm for space-time division of projector light rays. Our prototype system has up to four layers, with each layer consisting of an row of 50 drops that can be generated at up to 60 Hz. The effective resolution of the display is 50x projector vertical-resolution x number of layers. We show how this water drop display can be used for text, videos, and interactive games.

/pdf/a-multi-layered-display-with-water-drops-1xsvbzpd6o.pdf

A multi-layered display with water drops

Capturing good videos outdoors can be challenging due to harsh lighting, unpredictable scene changes, and most relevant to this work, dynamic weather. Particulate weather, such as rain and snow, creates complex flickering effects that are irritating to people and confusing to vision algorithms. Although each raindrop or snowflake only affects a small number of pixels, collections of them have predictable global spatio-temporal properties. In this paper, we formulate a model of these global dynamic weather frequencies. To begin, we derive a physical model of raindrops and snowflakes that is used to determine the general shape and brightness of a single streak. This streak model is combined with the statistical properties of rain and snow, to determine how they effect the spatio-temporal frequencies of an image sequence. Once detected, these frequencies can then be suppressed. At a small scale, many things appear the same as rain and snow, but by treating them as global phenomena, we achieve better performance than with just a local analysis. We show the effectiveness of removal on a variety of complex video sequences.

https://hal.inria.fr/inria-00264716/document

Spatio-Temporal Frequency Analysis for Removing Rain and Snow from Videos

In this paper, we present an approach to color image understanding that can be used to segment and analyze surfaces with color variations due to highlights and shading. We begin with a theory that relates the reflected light from dielectric materials, such as plastic, to fundamental physical reflection processes, and describes the color of the reflected light as a linear combination of the color of the light due to surface reflection (highlights) and body reflection (object color). This theory is used in an algorithm that separates a color image into two parts: an image of just the highlights, and the original image with the highlights removed. In the past, we have applied this method to hand-segmented images. The current paper shows how to perform automatic segmentation method by applying this theory in stages to identify the object and highlight colors. The result is a combination of segmentation and reflection analysis that is better than traditional heuristic segmentation methods (such as histogram thresholding), and provides important physical information about the surface geometry and material properties at the same time. We also show the importance of modeling the camera properties for this kind of quantitative analysis of color. This line of research cRn lead to physics-based image segmentation methods that are both more reliable and more useful than traditional segmentation methods.

/pdf/image-segmentation-and-reflection-analysis-through-color-2ousxnkhgo.pdf

Takeo Kanade

Papers

Spatially directed guidance of stem cell population migration by immobilized patterns of growth factors

3-D vision techniques for autonomous vehicles

A multi-layered display with water drops

Spatio-Temporal Frequency Analysis for Removing Rain and Snow from Videos

Image Segmentation and Reflection Analysis Through Color