ACTOR RETRIEVAL SYSTEM BASED ON KERNELS ON BAGS OF BAGS

International journal of scientific research, 2015

For a specific set of features selected for representing videos, the performance of a content-based video retrieval system depends critically on the similarity or dissimilarity measures used. In this paper, we propose a kernel approach to improve the retrieval performance of content-based video retrieval systems namely determinant kernel. The input of this kernel is the dot product of feature matrices that extracted from shot visual information. Due to the variation in the number of each shot frames, the size of feature matrices are different and so the result of dot product become a non-square matrices. Almost all available techniques use summarizing methods to equalize the size of the matrices which lead to loss some parts of information. To solve this problem, we present a non-square determinant kernel based on Radic's definition. We evaluate the performance of the derived Kernels by retrieving video shots of news and speaking videos. Experimental results confirm the effectiv...

2010

In this article, we propose the Peersommender architecture, which is the set of applications that provide personalized content to users according to their annotations produced when watching multimedia items. As opposite to hierarchical authoring, which is metadata created by experts to describe content in an organized, structured and impartial manner, peer-level annotations are highly personal because they are created by consumers, and this feature can be used to infer relevant content that is of interest to the user. Particularly, we propose a movie recommender system that explores a user profile with automatic augmentation, which is based on annotations produced by the user in the past. By combining tags, faces of interest and ratings with usual hierarchical metadata, we are able to predict ratings for new movies based on an enhanced hybrid approach for content filtering. Our evaluation was executed over a large scale dataset containing real users, and it shows good results when compared to other techniques.

International Journal of Data Mining, Modelling and Management, 2011

The subject of this paper is the kernel method, which offers an elegant solution to improve the effectiveness and the efficiency of the learning process in an Image Retrieval System. By simulating learning in high-dimensional feature spaces while working with the original low-dimensional input data, kernel methods provide a way for obtaining nonlinear decision boundaries from algorithms previously restricted to handling only linearly separable image collections. The aim of this paper is to summarize results that have been obtained by using the kernel method in Image Retrieval Systems. First, the paper introduces the kernel method together with some of its useful properties, and several kernel types used in fields like Image Retrieval, Text Mining, Machine Learning, and Computational Biology. Then, the paper focuses on applications and research questions involving the kernel method, such as kernel-based learning, kernel selection, and feature selection methods.

In this paper, Content Based Video Retrieval Systems performance is analysed and compared for three different types of feature vectors. These types of features are generated using three different algorithms; Block Truncation Coding (BTC) extended for colors, Kekre's Fast Codebook Generation (KFCG) algorithm and Gabor filters. The feature vectors are extracted from multiple frames instead of using only key frames or all frames from the videos. The performance of each type of feature is analysed by comparing the results obtained by two different techniques; Euclidean Distance and Support Vector Machine (SVM). Although a significant number of researchers have expressed dissatisfaction to use image as a query for video retrieval systems, the techniques and features used here provide enhanced and higher retrieval results while using images from the videos. Apart from higher efficiency, complexity has also been reduced as it is not required to find key frames for all the shots. The system is evaluated using a database of 1000 videos consisting of 20 different categories. Performance achieved using BTC features calculated from color components is compared with that achieved using Gabor features and with KFCG features. These performances are compared again with the performances obtained from systems using SVM and the systems without using SVM.

IEEE Transactions on Consumer Electronics, 2000

Automatic face indexing is an important technique for realizing actor-based video services in an IPTV environment. This paper proposes a novel face indexing system that takes advantages of the internet connection of an STB to construct a FR engine that is equipped with a high number of training face images. In addition, we use a face clustering technique to obtain multiple face images of the same subject from a sequence of video frames. The clustered face images are combined using a weighted feature fusion scheme, resulting in a considerable enhancement in face indexing accuracy. The effectiveness of the proposed system is validated using more than 300,000 video frames, collected from five video clips containing drama or movie content. The experimental results show that the proposed system can achieve a face annotation accuracy that is feasible for practical applications. 1

International Journal of Innovative Technology and Exploring Engineering, 2019

Information retrieval is one of the important areas of research with highest scope for data mining combined with machine learning. The proposed research focus on visual information retrieval by applying machine learning techniques. The usage of multimedia data such as text, images, videos are abundantly increasing day by day in this smart era. Also the need for information classification and retrieval are getting exponential demands to fulfill the research and end user requirements. The tech giants are conducting their researches to develop efficient retrieval systems for videos. Video retrieval is considered to be the toughest and challenging research in the recent times. Due to large storage space, lengthy play time, multiple sequence of frames, spatial temporal challenges, lack of visual relevancy, less hardware and processing support. The proposed visual information retrieval has got higher scope of research with the above listed problems.

2008

The actual generation of video search engines offers low-level abstractions of the data while users seek for high-level semantics. The main challenge in video retrieval remains bridging the semantic gap. Thus, the effectiveness of video retrieval is based on the result of the interaction between query selection and a goal-oriented human user. The system exploits the human capability for rapidly scanning imagery augmenting it with an active learning loop, which tries to always present the most relevant material based on the current information. We describe in this paper, a machine learning system for interactive video retrieval. The core of this system is a kernel-based SVM classifier. The video retrieval uses the core as an active learning classifier. We perform an experiment against the 2005 NIST TRECVID benchmark in the high-level task.

Principal component analysis (PCA) is an important tool in many areas including data reduction and interpretation, information retrieval, image processing, and so on. Kernel PCA has recently been proposed as a nonlinear extension of the popular PCA. The basic idea is to first map the input space into a feature space via a nonlinear map and then compute the principal components in that feature space. This paper illustrates the potential of kernel PCA for dimensionality reduction and feature extraction in multimedia retrieval. By the use of Gaussian kernels, the principal components were computed in the feature space of an image data set and they are used as new dimensions to approximate image features. Extensive experimental results show that kernel PCA performs better than linear PCA with respect to the retrieval quality as well as the retrieval precision in content-based image retrievals.

Similarity Search and Applications 12th International Conference, SISAP 2019, 2019

Since the 1970's the Content-Based Image Indexing and Retrieval (CBIR) has been an active area. Nowadays, the rapid increase of video data has paved the way to the advancement of the technologies in many different communities for the creation of Content-Based Video Indexing and Retrieval (CBVIR). However, greater attention needs to be devoted to the development of effective tools for video search and browse. In this paper, we present Visione, a system for large-scale video retrieval. The system integrates several content-based analysis and retrieval modules , including a keywords search, a spatial object-based search, and a visual similarity search. From the tests carried out by users when they needed to find as many correct examples as possible, the similarity search proved to be the most promising option. Our implementation is based on state-of-the-art deep learning approaches for content analysis and leverages highly efficient indexing techniques to ensure scalability. Specifically, we encode all the visual and textual descriptors extracted from the videos into (surrogate) textual representations that are then efficiently indexed and searched using an off-the-shelf text search engine using similarity functions.

IEEE Access, 2019

Recent advances in the film industry have given rise to exponential growth in movie/drama production and adaptation of the Big Data concept. Automatic identification and classification of movie characters has received tremendous attention from researchers due to its applications in video semantic analysis, video summarization and personalized video retrieval for which several methods have been recently presented. However, these methods cannot detect main characters properly due to their variation in pose and style in different scenes of a movie. To address this problem we present DeepStar, a novel framework for starring character identification based on deep high-level robust features. The proposed framework is threefold: the extraction of shots with clear faces from the input video; face clustering using discriminative deep features; and the occurrence matrix generation, helping in the selection of starring characters. The promising results obtained using representative Hollywood movies demonstrate the effectiveness of our method in detecting starring characters over state-of-the-art methods.