Search

We propose a framework for automated classification of Advertisement Images, using not just Visual features but also Textual cues extracted from embedded text. Our approach takes inspiration from the assumption that Ad images contain meaningful textual content, that can provide discriminative semantic interpretetion, and can thus aid in classifcation tasks. To this end, we develop a framework using off-the-shelf components, and demonstrate the effectiveness of Textual cues in semantic Classfication tasks., Comment: Accepted in CVPR 2018 Workshop: Towards Automatic Understanding of Visual Advertisements (ADS)

In this work we introduce a cross modal image retrieval system that allows both text and sketch as input modalities for the query. A cross-modal deep network architecture is formulated to jointly model the sketch and text input modalities as well as the the image output modality, learning a common embedding between text and images and between sketches and images. In addition, an attention model is used to selectively focus the attention on the different objects of the image, allowing for retrieval with multiple objects in the query. Experiments show that the proposed method performs the best in both single and multiple object image retrieval in standard datasets., Comment: Accepted at ICPR 2018

We consider network coding for a noiseless broadcast channel where each receiver demands a subset of messages available at the transmitter and is equipped with noisy side information in the form an erroneous version of the message symbols it demands. We view the message symbols as elements from a finite field and assume that the number of symbol errors in the noisy side information is upper bounded by a known constant. This communication problem, which we refer to as 'broadcasting with noisy side information' (BNSI), has applications in the re-transmission phase of downlink networks. We derive a necessary and sufficient condition for a linear coding scheme to satisfy the demands of all the receivers in a given BNSI network, and show that syndrome decoding can be used at the receivers to decode the demanded messages from the received codeword and the available noisy side information. We represent BNSI problems as bipartite graphs, and using this representation, classify the family of problems where linear coding provides bandwidth savings compared to uncoded transmission. We provide a simple algorithm to determine if a given BNSI network belongs to this family of problems, i.e., to identify if linear coding provides an advantage over uncoded transmission for the given BNSI problem. We provide lower bounds and upper bounds on the optimal codelength and constructions of linear coding schemes based on linear error correcting codes. For any given BNSI problem, we construct an equivalent index coding problem. A linear code is a valid scheme for a BNSI problem if and only if it is valid for the constructed index coding problem., Comment: 22 pages, 4 figures

Offline signature verification is one of the most challenging tasks in biometrics and document forensics. Unlike other verification problems, it needs to model minute but critical details between genuine and forged signatures, because a skilled falsification might often resembles the real signature with small deformation. This verification task is even harder in writer independent scenarios which is undeniably fiscal for realistic cases. In this paper, we model an offline writer independent signature verification task with a convolutional Siamese network. Siamese networks are twin networks with shared weights, which can be trained to learn a feature space where similar observations are placed in proximity. This is achieved by exposing the network to a pair of similar and dissimilar observations and minimizing the Euclidean distance between similar pairs while simultaneously maximizing it between dissimilar pairs. Experiments conducted on cross-domain datasets emphasize the capability of our network to model forgery in different languages (scripts) and handwriting styles. Moreover, our designed Siamese network, named SigNet, exceeds the state-of-the-art results on most of the benchmark signature datasets, which paves the way for further research in this direction.

This paper proposes a region based convolutional neural network for segmentation-free word spotting. Our net- work takes as input an image and a set of word candidate bound- ing boxes and embeds all bounding boxes into an embedding space, where word spotting can be casted as a simple nearest neighbour search between the query representation and each of the candidate bounding boxes. We make use of PHOC embedding as it has previously achieved significant success in segmentation- based word spotting. Word candidates are generated using a simple procedure based on grouping connected components using some spatial constraints. Experiments show that R-PHOC which operates on images directly can improve the current state-of- the-art in the standard GW dataset and performs as good as PHOCNET in some cases designed for segmentation based word spotting., Comment: Accepted in ICDAR'2017

In this paper we propose an approach to lexicon-free recognition of text in scene images. Our approach relies on a LSTM-based soft visual attention model learned from convolutional features. A set of feature vectors are derived from an intermediate convolutional layer corresponding to different areas of the image. This permits encoding of spatial information into the image representation. In this way, the framework is able to learn how to selectively focus on different parts of the image. At every time step the recognizer emits one character using a weighted combination of the convolutional feature vectors according to the learned attention model. Training can be done end-to-end using only word level annotations. In addition, we show that modifying the beam search algorithm by integrating an explicit language model leads to significantly better recognition results. We validate the performance of our approach on standard SVT and ICDAR'03 scene text datasets, showing state-of-the-art performance in unconstrained text recognition.

We apply the Dirac-Bergmann algorithm for the analysis of constraints to gauge theories defined on spherically symmetric black hole backgrounds. We find that the constraints for a given theory are modified on such spacetimes through the presence of additional contributions from the horizon. As a concrete example, we consider the Maxwell field on a black hole background, and determine the role of the horizon contributions on the dynamics of the theory., Comment: 23 pp

We construct a simple algorithm to derive number density of spin 1/2 particles created in spatially flat FLRW spacetimes and resulting renormalized energy-momentum tensor within the framework of adiabatic regularization. Physical quantities thus found are in agreement with the known results. This formalism can be considered as an appropriate extension of the techniques originally introduced for scalar fields, applicable to fermions in curved space. We apply this formalism to compute the particle number density and the renormalized energy density and pressure analytically (wherever possible) and numerically, in two interesting cosmological scenarios: a de Sitter spacetime and a radiation dominated universe. Results prove the efficiency of the methodology presented here., Comment: 12 pages, 10 figures, version accepted for publication in Physical Review D. arXiv admin note: substantial text overlap with arXiv:1506.06909

Within the framework of adiabatic regularization, we present a simple formalism to calculate number density and renormalized energy-momentum density of spin 1/2 particles in spatially flat FLRW spacetimes using an appropriate WKB ansatz for the adiabatic expansion for the field modes. The conformal and axial anomalies thus found are in exact agreement with those obtained from other renormalization methods. This formalism can be considered as an appropriate extension of the techniques originally introduced for scalar fields, applicable to fermions in curved space., Comment: 7 pages, accepted for publication in Phys. Rev. D

In this paper we propose a segmentation-free query by string word spotting method. Both the documents and query strings are encoded using a recently proposed word representa- tion that projects images and strings into a common atribute space based on a pyramidal histogram of characters(PHOC). These attribute models are learned using linear SVMs over the Fisher Vector representation of the images along with the PHOC labels of the corresponding strings. In order to search through the whole page, document regions are indexed per character bi- gram using a similar attribute representation. On top of that, we propose an integral image representation of the document using a simplified version of the attribute model for efficient computation. Finally we introduce a re-ranking step in order to boost retrieval performance. We show state-of-the-art results for segmentation-free query by string word spotting in single-writer and multi-writer standard datasets, Comment: To be published in ICDAR2015