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

We propose PALM-a portable sensor-augmented vision system for large-scene modeling. The system solves the problem of recovering large structures in arbitrary scenes from video streams taken by a sensor-augmented camera. Central to the solution method is the use of multiple constraints derived from GPS measurements, camera orientation sensor readings, and image features. The knowledge of camera orientation enhances computational efficiency by making a linear formulation of perspective ray constraints possible. The overall shape is constructed by merging smaller shape segments. Shape merging errors are minimized using the concept of shape hierarchy, which is realized through a "landmarking" technique. The features of the system include its use of a small number of images and feature points, its portability, and its low cost interface for synchronizing sensor measurements with the video stream. Example reconstructions of a football stadium and two large buildings are presented and these results are compared with the ground truth.

PALM: portable sensor-augmented vision system for large-scene modeling

Topics in Cryptology - CT-RSA 2009

We present appearance-based Virtual view generation which allows viewers to fly through a real dynamic scene. The scene is captured by synchronized multiple cameras. Arbitrary views are generated by interpolating two original camera-view images near the given viewpoint. The quality of the generated synthetic view is determined by the precision, consistency and density of correspondences between the two images. All or most of previous work that uses interpolation extracts the correspondences from these two images. However, not only is it difficult to do so reliably (the task requires a good stereo algorithm), but also the two images alone sometimes do not have enough information, due to problems such as occlusion. Instead, we take advantage of the fact that we have many views, from which we can extract much more reliable and comprehensive 3D geometry of the scene as a 3D model. The dense and precise correspondences between the two images, to be used for interpolation, are derived from this constructed 3D model. Our method of 3D modeling from multiple images uses the Multiple Baseline Stereo method and Shape from Silhouette method.

Appearance-Based Virtual-View Generation for Fly Through in a Real Dynamic Scene

This paper describes a system which can perform full 3-D pose estimation of a single arbitrarily shaped, rigid object at rates up to 10 Hz. A triangular mesh model of the object to be tracked is generated offline using conventional range sensors. Real-time range data of the object is sensed by the CMU high speed VLSI range sensor. Pose estimation is performed by registering the real-time range data to the triangular mesh model using an enhanced implementation of the Iterative Closest Point (ICP) Algorithm introduced by Besl and McKay (1992). The method does not require explicit feature extraction or specification of correspondence. Pose estimation accuracies of the order of 1% of the object size in translation, and 1 degree in rotation have been measured. >

Real-time 3-D pose estimation using a high-speed range sensor

Here we accelerate computations that reduce CT metal artifacts by observing that metal projects to only a fraction of the X-ray detectors in each projection, and thus computations should focus on these metal “mask” regions. We propose that the penalized maximum likelihood optimization method for artifact reduction needs to be solved only within the metal mask, using the remaining non-mask regions as a constraint; we show that our approach leads to a 10x speedup.

/pdf/fast-streaking-artifact-reduction-in-ct-using-constrained-tkxcjsovej.pdf

Takeo Kanade

Papers

PALM: portable sensor-augmented vision system for large-scene modeling

Topics in Cryptology - CT-RSA 2009

Appearance-Based Virtual-View Generation for Fly Through in a Real Dynamic Scene

Real-time 3-D pose estimation using a high-speed range sensor

Fast Streaking Artifact Reduction in CT Using Constrained Optimization in Metal Masks