Savas Ozkan

Performance Analysis of State-of-the-Art Representation Methods for Geographical Image Retrieval and Categorization

This letter studies the performance of various image representation schemes used for image search problems for the purpose of geographic image retrieval from satellite imagery. We compare the most widely adopted method of the bag-of-words (BoW) approach with the more recently introduced vector of locally aggregated descriptors (VLAD) and its more compact binary version product quantized VLAD (VLAD-PQ). We show with the experiments on a publicly available 21-class land-use/land-cover data set that the VLAD-based representation outperforms BoW at the cost of increased query time, but the more compact VLAD-PQ representation achieves very similar performance as VLAD without the increased time requirement.

Visual Group Binary Signature for Video Copy Detection

Need for automatic video copy detection is increased with the recent technical developments in the internet technologies and video recording. Even though image-based techniques with bag-of-word kind of representations are accepted as the best solution because of robustness and speed, they discard the convenient geometric relation which exists among interest points. In this work, we propose a novel geometric relation which computes a binary signature leveraging existence and non-existence of interest points in the neighborhood area. The experimental results on TRECVID 2009 content-based video copy detection dataset show that combination of our method with recently proposed quantization-based indexing and weak geometric consistency schemes outperforms classical representations.

Towards Category Based Large-Scale Image Retrieval Using Transductive Support Vector Machines

In this study, we use transductive learning and binary hierarchical trees to create compact binary hashing codes for large-scale image retrieval applications. We create multiple hierarchical trees based on the separability of the visual object classes by random selection, and the transductive support vector machine (TSVM) classifier is used to separate both the labeled and unlabeled data samples at each node of the binary hierarchical trees (BHTs). Then the separating hyperplanes returned by TSVM are used to create binary codes. We propose a novel TSVM method that is more robust to the noisy labels by interchanging the classical Hinge loss with the robust Ramp loss. Stochastic gradient based solver is used to learn TSVM classifier to ensure that the method scales well with large-scale data sets. The proposed method improves the Euclidean distance metric and achieves comparable results to the state-of-art on CIFAR10 and MNIST data sets and significantly outperforms the state-of-art hashing methods on NUS-WIDE dataset.

Enhanced spatio-temporal video copy detection by combining trajectory and spatial consistency

The recent improvements on internet technologies and video coding techniques cause an increase in copyright infringements especially for video. Frequently, image-based approaches appear as an essential solution due to the fact that joint usage of quantization-based indexing and weak geometric consistency stages give a capability to compare duplicate videos quickly. However, exploiting purely spatial content ignores the temporal variation of video. In this work, we propose a system that combines the state-of-the-art quantization-based indexing scheme with a novel trajectory-based geometric consistency on spatio-temporal features. This combination improves duplicate video matching task significantly. Briefly, spatial mean and variance of the trajectories are incorporated to establish a weak geometric consistency among pair of frames. To show the success of the proposed method, content-based video copy detection field is selected and TRECVID 2009 dataset is utilized. The experimental results show that constituting trajectory-based consistency on corresponding feature pairs outperforms the performances of merely utilizing spatiotemporal signature and visual signature with enhanced weak geometric consistency.

Multimodal concept detection in broadcast media: KavTan

Concept detection stands as an important problem for efficient indexing and retrieval in large video archives. In this work, the KavTan System, which performs high-level semantic classification in one of the largest TV archives of Turkey, is presented. In this system, concept detection is performed using generalized visual and audio concept detection modules that are supported by video text detection, audio keyword spotting and specialized audio-visual semantic detection components. The performance of the presented framework was assessed objectively over a wide range of semantic concepts (5 high-level, 14 visual, 9 audio, 2 supplementary) by using a significant amount of precisely labeled ground truth data. KavTan System achieves successful high-level concept detection performance in unconstrained TV broadcast by efficiently utilizing multimodal information that is systematically extracted from both spatial and temporal extent of multimedia data.

Relevance feedback for semantic classification: A comparative study

Immense increase in the number of multimedia content accessible from television and internet with the help developing technologies reveals efficient supervision and classification of such content as a problem. Relevance feedback is a technique which relies on evaluation of retrieval results by humans and enables reduce the semantic gap between ideas and low level representations. Content based high level classification system may employ relevance feedback for improved retrieval performance. In this paper, different relevance feedback algorithms, which can be utilized to increase generalized semantic classification performance, are discussed and compared inside an experimental framework. Some improvements are also proposed over obtained results.

Noise reduction on hyperspectral imagery using spectral unmixing and class-labels

Noise reduction on hyperspectral imagery is a critical step for the success of other applications that use this type of data. In this paper, we propose a novel approach to reduce the noise on hyperspectral data that might occur due to various factors. Since the proposed method exploits class-labels of data, it can be categorized as a semi-supervised method. First, our approach computes the mean spectral signatures of data using their spatial coherence and class-labels, then robust pure material signatures are estimated with different spectral unmixing methods. Later, these signatures are analyzed for the noise reduction. Tests are conducted on Indian Pines dataset under different noise characteristics. The experimental results show that our proposed method improves PSNR scores compared to baseline methods that use either spectral unmixing or class-labels. Furthermore, noticeable improvements on computation complexity are observed.

Analysis of Turkish parliament records in terms of party coherence

In natural language processing and text mining, highly successful applications are developed with the recently introduced techniques. Particularly, noticeable performance increases are achieved on countless applications by using word embedding method. In this paper, we propose a novel text mining method based on word embedding and Fisher vector. The automatic analysis of political records is selected as a special application. Second novelty of this paper is the definition of party coherence and its objective evaluation with the proposed method. Experiments conducted on Turkish parliament records clearly show the apparent coherence of the party speeches and the differentiation with other parties. In the light of the experimental results, our proposed method can be utilized in many other domains including the political analysis.

Large-scale image retrieval using transductive support vector machines

Abstract In this paper, we propose a new method for large-scale image retrieval by using binary hierarchical trees and transductive support vector machines (TSVMs). We create multiple hierarchical trees based on the separability of the visual object classes, and TSVM classifier is used to find the hyperplane that best separates both the labeled and unlabeled data samples at each node of the binary hierarchical trees (BHTs). Then the separating hyperplanes returned by TSVM are used to create binary codes or to reduce the dimension. We propose a novel TSVM method that is more robust to the noisy labels by interchanging the classical Hinge loss with the robust Ramp loss. Stochastic gradient based solver is used to learn TSVM classifier to ensure that the method scales well with large-scale data sets. The proposed method significantly improves the Euclidean distance metric and achieves comparable results to the state-of-the-art on CIFAR10 and MNIST data sets, and significantly outperforms the state-of-the-art hashing methods on more challenging ImageCLEF 2013, NUS-WIDE, and CIFAR100 data sets.

Feasible local content representation for image-in-video search on large-video collection

In this paper, we tackle content-based image-in-video search on large video archive. Particularly in this problem, the computation cost of a proposed method is another important parameter that should be considered alongside of the success rate achieved due to the high volume of data. With our proposed method in this paper, content representation of multimedia data is achieved quite fast compared to other local approaches. Additionally, approximately %16 improvement is introduced at the success rate with respect to the method that uses global approach for content representation on Stanford I2V dataset.