Abstract: This note introduces a visual attention model of text localization in real-world scenes. The core of the model built upon the proto-object concept is discussed. It is shown how such dynamic mid-level representation of the scene can be derived in the framework of an action-perception loop engaging salience, text information value computation, and eye guidance mechanisms.
Preliminary results that compare model generated scanpaths with those eye-tracked from human subjects are presented.

Abstract: Despite the “paperless office” ideal that started in the decade of the seventies, businesses still strive against an increasing amount of paper documentation. Companies still receive huge amounts of paper documentation that need to be analyzed and processed, mostly in a manual way. A solution for this task consists in, first, automatically scanning the incoming documents. Then, document images can be analyzed and information can be extracted from the data. Documents can also be automatically dispatched to the appropriate workflows, used to retrieve similar documents in the dataset to transfer information, etc.
Due to the nature of this “digital mailroom”, we need document representation methods to be general, i.e., able to cope with very different types of documents. We need the methods to be sound, i.e., able to cope with unexpected types of documents, noise, etc. And, we need to methods to be scalable, i.e., able to cope with thousands or millions of documents that need to be processed, stored, and consulted. Unfortunately, current techniques of document representation, classification and retrieval are not apt for this digital mailroom framework, since they do not fulfill some or all of these requirements.
Through this thesis we focus on the problem of document representation aimed at classification and retrieval tasks under this digital mailroom framework. We first propose a novel document representation based on runlength histograms, and extend it to cope with more complex documents such as multiple-page documents, or documents that contain more sources of information such as extracted OCR text. Then we focus on the scalability requirements and propose a novel binarization method which we dubbed PCAE, as well as two general asymmetric distances between binary embeddings that can significantly improve the retrieval results at a minimal extra computational cost. Finally, we note the importance of supervised learning when performing large-scale retrieval, and study several approaches that can significantly boost the results at no extra cost at query time.

Abstract: In this paper we describe a framework to learn a model of shape variability in a set of patterns. The framework is based on the Active Appearance Model (AAM) and permits to combine shape deformations with appearance variability. We have used two modifications of the Blurred Shape Model (BSM) descriptor as basic shape and appearance features to learn the model. These modifications permit to overcome the rigidity of the original BSM, adapting it to the deformations of the shape to be represented. We have applied this framework to representation and classification of handwritten digits and symbols. We show that results of the proposed methodology outperform the original BSM approach.

Abstract: In this paper we propose an unsupervised segmentation-free method for word spotting in document images.
Documents are represented with a grid of HOG descriptors, and a sliding window approach is used to locate the document regions that are most similar to the query. We use the exemplar SVM framework to produce a better representation of the query in an unsupervised way. Finally, the document descriptors are precomputed and compressed with Product Quantization. This offers two advantages: first, a large number of documents can be kept in RAM memory at the same time. Second, the sliding window becomes significantly faster since distances between quantized HOG descriptors can be precomputed. Our results significantly outperform other segmentation-free methods in the literature, both in accuracy and in speed and memory usage.

Abstract: Representing a graph with a feature vector is a common way of making statistical machine learning algorithms applicable to the domain of graphs. Such a transition from graphs to vectors is known as graphembedding. A key issue in graphembedding is to select a proper set of features in order to make the vectorial representation of graphs as strong and discriminative as possible. In this article, we propose features that are constructed out of frequencies of node label representatives. We first build a large set of features and then select the most discriminative ones according to different ranking criteria and feature transformation algorithms. On different classification tasks, we experimentally show that only a small significant subset of these features is needed to achieve the same classification rates as competing to state-of-the-art methods.

Abstract: This paper addresses the problem of learning similarity-preserving binary codes for efficient similarity search in large-scale image collections. We formulate this problem in terms of finding a rotation of zero-centered data so as to minimize the quantization error of mapping this data to the vertices of a zero-centered binary hypercube, and propose a simple and efficient alternating minimization algorithm to accomplish this task. This algorithm, dubbed iterative quantization (ITQ), has connections to multi-class spectral clustering and to the orthogonal Procrustes problem, and it can be used both with unsupervised data embeddings such as PCA and supervised embeddings such as canonical correlation analysis (CCA). The resulting binary codes significantly outperform several other state-of-the-art methods. We also show that further performance improvements can result from transforming the data with a nonlinear kernel mapping prior to PCA or CCA. Finally, we demonstrate an application of ITQ to learning binary attributes or “classemes” on the ImageNet dataset.

Abstract: The combination of multiple features or views when representing documents or other kinds of objects usually leads to improved results in classification (and retrieval) tasks. Most systems assume that those views will be available both at training and test time. However, some views may be too `expensive' to be available at test time. In this paper, we consider the use of Canonical Correlation Analysis to leverage `expensive' views that are available only at training time. Experimental results show that this information may significantly improve the results in a classification task.