https://www2.econ.iastate.edu/tesfatsi/DeepLearningInNeuralNetworksOverview.JSchmidhuber2015.pdf

Deep learning in neural networks

In this work we address the task of semantic image segmentation with Deep Learning and make three main contributions that are experimentally shown to have substantial practical merit. First , we highlight convolution with upsampled filters, or ‘atrous convolution’, as a powerful tool in dense prediction tasks. Atrous convolution allows us to explicitly control the resolution at which feature responses are computed within Deep Convolutional Neural Networks. It also allows us to effectively enlarge the field of view of filters to incorporate larger context without increasing the number of parameters or the amount of computation. Second , we propose atrous spatial pyramid pooling (ASPP) to robustly segment objects at multiple scales. ASPP probes an incoming convolutional feature layer with filters at multiple sampling rates and effective fields-of-views, thus capturing objects as well as image context at multiple scales. Third , we improve the localization of object boundaries by combining methods from DCNNs and probabilistic graphical models. The commonly deployed combination of max-pooling and downsampling in DCNNs achieves invariance but has a toll on localization accuracy. We overcome this by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF), which is shown both qualitatively and quantitatively to improve localization performance. Our proposed “DeepLab” system sets the new state-of-art at the PASCAL VOC-2012 semantic image segmentation task, reaching 79.7 percent mIOU in the test set, and advances the results on three other datasets: PASCAL-Context, PASCAL-Person-Part, and Cityscapes. All of our code is made publicly available online.

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs. However, for many tasks, paired training data will not be available. We present an approach for learning to translate an image from a source domain X to a target domain Y in the absence of paired examples. Our goal is to learn a mapping G : X → Y such that the distribution of images from G(X) is indistinguishable from the distribution Y using an adversarial loss. Because this mapping is highly under-constrained, we couple it with an inverse mapping F : Y → X and introduce a cycle consistency loss to push F(G(X)) ≈ X (and vice versa). Qualitative results are presented on several tasks where paired training data does not exist, including collection style transfer, object transfiguration, season transfer, photo enhancement, etc. Quantitative comparisons against several prior methods demonstrate the superiority of our approach.

Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

In this work we address the task of semantic image segmentation with Deep Learning and make three main contributions that are experimentally shown to have substantial practical merit. First, we highlight convolution with upsampled filters, or 'atrous convolution', as a powerful tool in dense prediction tasks. Atrous convolution allows us to explicitly control the resolution at which feature responses are computed within Deep Convolutional Neural Networks. It also allows us to effectively enlarge the field of view of filters to incorporate larger context without increasing the number of parameters or the amount of computation. Second, we propose atrous spatial pyramid pooling (ASPP) to robustly segment objects at multiple scales. ASPP probes an incoming convolutional feature layer with filters at multiple sampling rates and effective fields-of-views, thus capturing objects as well as image context at multiple scales. Third, we improve the localization of object boundaries by combining methods from DCNNs and probabilistic graphical models. The commonly deployed combination of max-pooling and downsampling in DCNNs achieves invariance but has a toll on localization accuracy. We overcome this by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF), which is shown both qualitatively and quantitatively to improve localization performance. Our proposed "DeepLab" system sets the new state-of-art at the PASCAL VOC-2012 semantic image segmentation task, reaching 79.7% mIOU in the test set, and advances the results on three other datasets: PASCAL-Context, PASCAL-Person-Part, and Cityscapes. All of our code is made publicly available online.

This paper presents a new representation and evaluation procedure of page segmentation algorithms and analyzes six widely-used layout analysis algorithms using the procedure. The method permits a detailed analysis of the behavior of page segmentation algorithms in terms of over- and undersegmentation at different layout levels, as well as determination of the geometric accuracy of the segmentation. The representation of document layouts relies on labeling each pixel according to its function in the overall segmentation, permitting pixel-accurate representation of layout information of arbitrary layouts and allowing background pixels to be classified as "dont care". Our representations can be encoded easily in standard color image formats like PNG, permitting easy interchange of segmentation results and ground truth.

Pixel-Accurate Representation and Evaluation of Page Segmentation in Document Images

Object recognition in X-ray images is an interesting application of machine vision that can help reduce the workload of human operators of X-ray scanners at security checkpoints. In this paper, we first present a comprehensive evaluation of image classification and object detection in X-ray images using standard local features in a BoW framework with (structural) SVMs. Then, we extend the features to utilize the extra information available in dual energy X-ray images. Finally, we propose a multi-view branch-and-bound algorithm for multi-view object detection. Through extensive experiments on three object categories, we show that the classification and detection performance substantially improves with the extended features and multiple views.

/pdf/object-recognition-in-multi-view-dual-energy-x-ray-images-3kqt029vqu.pdf

Object Recognition in Multi-View Dual Energy X-ray Images.

The detection of counterfeit in printed documents is currently based mainly on built-in security features or on human expertise. We propose a classification system that supports non-expert users to distinguish original documents from PC-made forgeries by analyzing the printing technique used. Each letter in a document is classified using a support vector machine that has been trained to distinguish laser from inkjet printouts. A color coded visualization helps the user to interpret the per-letter classification results.

https://www.dfki.de/web/forschung/iwi/publikationen/renameFileForDownload?filename=CHLMeiTMBPrintingTechClass.pdf&file_id=uploads_514

Printing Technique Classification for Document Counterfeit Detection

Most methods for document image retrieval rely solely on text information to find similar documents. This paper describes a way to use layout information for document image retrieval instead. A new class of distance measures is introduced for documents with Manhattan layouts, based on a two-step procedure: First, the distances between the blocks of two layouts are calculated. Then, the blocks of one layout are assigned to the blocks of the other layout in a matching step. Different block distances and matching methods are compared and evaluated using the publicly available MARG database. On this dataset, the layout type can be determined successfully in 92.6% of the cases using the best distance measure in a nearest neighbor classifier. The experiments show that the best distance measure for this task is the overlapping area combined with the Manhattan distance of the corner points as block distance together with the minimum weight edge cover matching.

/pdf/distance-measures-for-layout-based-document-image-retrieval-59ek4ieorf.pdf

Distance measures for layout-based document image retrieval

Many document analysis and OCR systems depend on precise identification of page rotation, as well as the reliable identification of text lines. This paper presents a new algorithm to address both problems. It uses a branch-and-bound approach to globally optimal line finding and simultaneously models the baseline and the descender line under a Gaussian error/robust least square model. Results of applying the algorithm to documents in the University of Washington Database 2 are presented.

Thomas M. Breuel

Papers

Pixel-Accurate Representation and Evaluation of Page Segmentation in Document Images

Object Recognition in Multi-View Dual Energy X-ray Images.

Printing Technique Classification for Document Counterfeit Detection

Distance measures for layout-based document image retrieval

Robust least-square-baseline finding using a branch and bound algorithm