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

: This paper presents the results of our experiments with a unique multiple-baseline stereo technique. This algorithm for producing precise, unambiguous depth maps from a set of multiple stereo pairs was developed by Okutomi and Kanade. Early versions of the algorithm were shown to perform well under controlled conditions in the Calibrated Imaging Laboratory (CIL). in this paper, the algorithm is further applied to complex outdoor scenes with variable lighting conditions and large depth ranges. While Okutomi and Kanade used stereo pairs acquired by moving a camera horizontally, we also investigated the use of stereo pairs taken by moving a camera in both horizontal and vertical directions. The use of stereo images with two orthogonal baseline orientations removes ambiguity and increases precision without the problems associated with the orientation of the features in a scene. We also show that the shapes of the sum of squared-difference (SSD) curve indicates the reliability of the match, and suggest a method to classify matches into various types and to improve estimates when a false match occurs. Finally, results are presented to show the effectiveness of this algorithm and the classification method.

Experiments in Multiple-Baseline Stereo

Mixed reality: where real and virtual worlds meet

In most autonomous driving applications, such as parking and commuting, a vehicle follows a previously taken route, or almost the same route. In this paper, we propose a method to localize a vehicle along a previously driven route using images. The proposed method consists of two stages: offline creation of a database, and online localization. In the offline stage, a database is created from images that are captured when the vehicle drives a route for the first time. The database consists of images, 3D positions of feature points estimated by structure-from-motion, and a topological graph. In the online stage, the method first identifies the database image that is most similar to the current image by topometric localization, which considers topological information on a metric scale. The vehicle poses are then estimated from the 3D-2D correspondences of matching feature points between the current image and the identified database image. In an experiment, we estimated vehicle poses using images captured in an indoor parking lot.

Takeo Kanade

Papers

Experiments in Multiple-Baseline Stereo

Mixed reality: where real and virtual worlds meet

Vehicle Localization along a Previously Driven Route Using Image Database.

Numerical Validation in Current Hardware Architectures

A theory of the origami world