Search

Multi-label classification problems usually occur in tasks related to information retrieval, like text and image annotation, and are receiving increasing attention from the machine learning and pattern recognition fields. One of the main issues under investigation is the development of classification algorithms capable of maximizing specific accuracy measures based on precision and recall. We focus on the widely used F measure, defined for binary, single-label problems as the weighted harmonic mean of precision and recall, and later extended to multi-label problems in three ways: macro-averaged, micro-averaged and instance-wise. In this paper we give a comprehensive survey of theoretical results and algorithms aimed at maximizing F measures. We subdivide it according to the two main existing approaches: empirical utility maximization, and decision-theoretic. Under the former approach, we also derive the optimal (Bayes) classifier at the population level for the instance-wise and micro-averaged F , extending recent results about the single-label F . In a companion paper we shall focus on the micro-averaged F measure, for which relatively fewer solutions exist, and shall develop novel maximization algorithms under both approaches. [ABSTRACT FROM AUTHOR]

Abstract: Intra-shape deformations complicate 3D shape recognition and therefore need proper modeling. Thereto, an isometric deformation model is used in this paper. The method proposed does not need explicit point correspondences for the comparison of 3D shapes. The geodesic distance matrix is used as an isometry-invariant shape representation. Two approaches are described to arrive at a sampling order invariant shape descriptor: the histogram of geodesic distance matrix values and the set of largest singular values of the geodesic distance matrix. Shape comparison is performed by comparison of the shape descriptors using the as dissimilarity measure. For object recognition, the results obtained demonstrate the singular value approach to outperform the histogram-based approach, as well as the state-of-the-art multidimensional scaling technique, the ICP baseline algorithm and other isometric deformation modeling methods found in literature. Using the TOSCA database, a rank-1 recognition rate of 100% is obtained for the identification scenario, while the verification experiments are characterized by a 1.58% equal error rate. External validation demonstrates that the singular value approach outperforms all other participants for the non-rigid object retrieval contests in SHREC 2010 as well as SHREC 2011. For 3D face recognition, the rank-1 recognition rate is 61.9% and the equal error rate is 11.8% on the BU-3DFE database. This decreased performance is attributed to the fact that the isometric deformation assumption only holds to a limited extent for facial expressions. This is also demonstrated in this paper. [Copyright &y& Elsevier]

Highlights • Important application areas such as signature extraction, ink or document aging, information retrieval from historical document images, paintings analysis and forensic analysis of documents are reviewed. • Challenges involved in the hyperspectral document image acquisition are discussed. • Challenges involved in the pre-processing of HS document images are discussed. • A detailed review and discussion of the commercial HSI systems for document image analysis is presented. • The survey will provide useful information to researchers and forensic scientists towards this exciting technology and help them to solve many challenging problems of document image analysis. Abstract Automatic image analysis is a crucial component of many intelligent systems designed for high-level understanding of documents. Most document image understanding systems are usually based on applying pattern recognition techniques to conventional three channel RGB images. Airborne and satellite based macro scale Hyperspectral Imaging (HSI) systems are well established for geosciences. Recently, owing to advancements in imaging speed and reduced camera costs, micro scale HSI systems are also gaining importance in ground based applications such as hyperspectral document image analysis. HSI is non destructive and offers new opportunities via measuring richer information along spectral dimension by imaging the document in contiguous bands across the electromagnetic spectrum. Hyperspectral document imaging has shown potential for solving many challenging problems of document image analysis including signature extraction, ink or document aging, information retrieval from historical document images, paintings and forensic analysis of documents. In this paper, we explore the potential of HSI for document image analysis and present a comprehensive review of the literature and future prospects. We highlight and discuss the challenges involved in the acquisition and processing of hyperspectral document images. A review of commercial HSI systems for document image analysis is also presented. [ABSTRACT FROM AUTHOR]

Highlights • We present a new framework of deep convolutional neural network architecture distillation, namely LightweightNet, for acceleration and compression. • We exploit the prior knowledge of pre-defined network architecture to guide the efficient design of acceleration/compression strategies, while not using pre-trained model. • The proposed framework consists of network parameter compression, network structure acceleration, and non-tensor layer improvement. • The proposed framework demonstrates a higher acceleration/compression rate than previous methods in experiments, including a large category handwritten Chinese character recognition task with state-of-the-art performance. Abstract In recent years, deep neural networks have achieved remarkable successes in many pattern recognition tasks. However, the high computational cost and large memory overhead hinder them from applications on resource-limited devices. To address this problem, many deep network acceleration and compression methods have been proposed. One group of methods adopt decomposition and pruning techniques to accelerate and compress a pre-trained model. Another group designs single compact unit to stack their own networks. These methods are subject to complicated training processes, or lack of generality and extensibility. In this paper, we propose a general framework of architecture distillation, namely LightweightNet, to accelerate and compress convolutional neural networks. Rather than compressing a pre-trained model, we directly construct the lightweight network based on a baseline network architecture. The LightweightNet, designed based on a comprehensive analysis of the network architecture, consists of network parameter compression, network structure acceleration, and non-tensor layer improvement. Specifically, we propose the strategy of low-dimensional features of fully-connected layers for substantial memory saving, and design multiple efficient compact blocks to distill convolutional layers of baseline network with accuracy-sensitive distillation rule for notable time saving. Finally, it can effectively reduce the computational cost and the model size by > 4 × with negligible accuracy loss. Benchmarks on MNIST, CIFAR-10, ImageNet and HCCR (handwritten Chinese character recognition) datasets demonstrate the advantages of the proposed framework in terms of speed, performance, storage and training process. In HCCR, our method even outperforms traditional handcrafted features-based classifiers in terms of speed and storage while maintaining state-of-the-art recognition performance. [ABSTRACT FROM AUTHOR]

In many visual classification tasks finding semantically meaningful regions has been confirmed as an effective solution. This paper aims to improve the performance of action classification in still images by introducing a discriminative region selection method. We observed that humans have certain periodic or symmetric pairs and they are critical for recognition. We also demonstrate that in action classification semantically meaningful regions are close to their periodic or symmetric parts and propose a model called a Generalized Symmetric Pair Model. By learning a max margin classifier, this method could identify regions around periodic or symmetric pairs without detection techniques. The method utilizes both the characteristics of actions and knowledge regarding periodism and symmetry to improve the popular bag-of-words (BoW) framework. We evaluate our method on five challenging action classification datasets. Experiments show that our method outperforms the state-of-the-art on four datasets. Qualitative visualization also demonstrate that the proposed method indeed identify semantically meaningful regions. [ABSTRACT FROM AUTHOR]

Information retrieval in historical documents has long consisted in spotting words. In this paper, we focus on graphical pattern spotting. Contrary to object detection and classification, where models of the object of interest may be trained, pattern spotting does not rely on any prior information regarding the query, nor predefined class of graphical objects. An offline sliding window approach may be suitable, provided that the challenge raised by high computational and storage costs is handled. We propose an unsupervised, segmentation-free approach that takes advantage of recent developments in computer vision to overcome these issues. We also investigate the use of new, compact descriptors for the data, namely the vectors of locally aggregated descriptors (VLAD) and Fisher Vectors, instead of the usual bag-of-visual-words approach. Results obtained on medieval manuscripts from the DocExplore project show that our approach achieves better retrieval results, with a better efficiency in terms of time/memory, compared to standard approaches. Experimentations show that VLAD and Fisher Vectors can be fruitfully used in the future for the description of historical documents. Additionally, we show that our system can be easily turned into a word spotting system with slight adaptation, and that it achieves results comparable to those recently published in ICDAR 2015 keyword spotting challenge. [ABSTRACT FROM AUTHOR]

Identifying individuals in photographs of animals collected over time is a non-invasive approach for ecological monitoring and conservation. This paper describes the design and use of Sloop, the first image retrieval system for individual animal identification incorporating crowd-sourced relevance feedback. Sloop׳s iterative retrieval strategy using hierarchical and aggregated matching and relevance feedback consistently improves deformation and correspondence-based approaches for individual identification across several species. Its crowdsourcing strategy is successful in utilizing relevance feedback on a large scale. Sloop is in operational use. The user experience and results are presented here to facilitate the creation of a community-based individual identification system for conservation planning. [ABSTRACT FROM AUTHOR]

Abstract: 2D cartoon plays an important role in many areas, but it requires effective methods to relieve manual labors. In this paper, we propose a heterogeneous cartoon gesture recognition method with applications. Firstly, heterogeneous features with different dimensions are assigned to express cartoon and human-subject images according to their characteristics. Then for recognition, we simultaneously integrate shared structure learning (SSL) and graph-based transductive learning into a joint framework to learn reliable classifiers on heterogeneous features. Provided with the framework, the similarities between cartoon and human-subject gestures can be quantitatively evaluated in a cross-feature manner. Extensive experiments on self-defined datasets have demonstrated the effectiveness of our method. Finally, applications illustrate the usages in various aspects of 2D cartoon industry. [Copyright &y& Elsevier]

Abstract: In this paper, we propose two general multiple-instance active learning (MIAL) methods, multiple-instance active learning with a simple margin strategy (S-MIAL) and multiple-instance active learning with fisher information (F-MIAL), and apply them to the active learning in localized content based image retrieval (LCBIR). S-MIAL considers the most ambiguous picture as the most valuable one, while F-MIAL utilizes the fisher information and analyzes the value of the unlabeled pictures by assigning different labels to them. In experiments, we will show their superior performances in LCBIR tasks. [Copyright &y& Elsevier]

Abstract: This paper proposes an automatic text-independent writer identification framework that integrates an industrial handwriting recognition system, which is used to perform an automatic segmentation of an online handwritten document at the character level. Subsequently, a fuzzy c-means approach is adopted to estimate statistical distributions of character prototypes on an alphabet basis. These distributions model the unique handwriting styles of the writers. The proposed system attained an accuracy of 99.2% when retrieved from a database of 120 writers. The only limitation is that a minimum length of text needs to be present in the document in order for sufficient accuracy to be achieved. We have found that this minimum length of text is about 160 characters or approximately equivalent to 3 lines of text. In addition, the discriminative power of different alphabets on the accuracy is also reported. [Copyright &y& Elsevier]