• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2407-2424
• /
• 2022

Recently, Transformer has made great progress in video retrieval tasks due to its high representation capability. For the structure of a Transformer, the cascaded self-attention modules are capable of capturing long-distance feature dependencies. However, the local feature details are likely to have deteriorated. In addition, increasing the depth of the structure is likely to produce learning bias in the learned features. In this paper, an improved Transformer structure named TransDCS (Transformer with Dynamic Convolution and Shortcut) is proposed. A Multi-head Conv-Self-Attention module is introduced to model the local dependencies and improve the efficiency of local features extraction. Meanwhile, the augmented shortcuts module based on a dual identity matrix is applied to enhance the conduction of input features, and mitigate the learning bias. The proposed model is tested on MSRVTT, LSMDC and Activity-Net benchmarks, and it surpasses all previous solutions for the video-text retrieval task. For example, on the LSMDC benchmark, a gain of about 2.3% MdR and 6.1% MnR is obtained over recently proposed multimodal-based methods.

• 한국HCI학회:학술대회논문집
• /
• 2006.02a
• /
• pp.926-930
• /
• 2006

Current image retrieval techniques have shortcomings that make it difficult to search for images based on a semantic understanding of what the image is about. Since an image is normally associated with multiple contexts (e.g. when and where a picture was taken,) the knowledge of these contexts can enhance the quantity of semantic understanding of an image. In this paper, we present a context-aware image retrieval system, which uses the context information to infer a kind of metadata for the captured images as well as images in different collections and databases. Experimental results show that using these kinds of information can not only significantly increase the retrieval accuracy in conventional content-based image retrieval systems but decrease the problems arise by manual annotation in text-based image retrieval systems as well.

• Asia pacific journal of information systems
• /
• v.1 no.1
• /
• pp.87-101
• /
• 1991

The primary purpose of this study was to investigate a robust search methodology that could be used in full-text information retrieval systems. A robust search methodology is one that can be easily used by a variety of users (particularly naive users) and it will give them comparable search performance regardless of their different expertise or interests In order to develop a possibly robust search methodology, a fully functional prototype of a fuzzy knowledge based information retrieval system was developed. Also, an experiment that used this prototype information retreival system was designed to investigate the performance of that search methodology over a small exploratory sample of user queries To probe the relatonships between the possibly robust search performance and the query organization using fuzzy inference logic, the search performance of a shallow query structure was analyzes. Consequently the following several noteworthy findings were obtained: 1) the hierachical(tree type) query structure might be a better query organization than the linear type query structure 2) comparing with the complex tree query structure, the simple tree query structure that has at most three levels of query might provide better search performance 3) the fuzzy search methodology that employs a proper levels of cut-off value might provide more efficient search performance than the boolean search methodology. Even though findings could not be statistically verified because the experiments were done using a single replication, it is worth noting however, that the research findings provided valuable information for developing a possibly robust search methodology in full-text information retrieval.

• Journal of radiological science and technology
• /
• v.28 no.1
• /
• pp.19-24
• /
• 2005

The purpose of this study was to design and implement a useful annotation-based Radiological image retrieval system to accurately determine on education and image information for Radiological technologists. For better retrieval performance based on large image databases, we presented an indexing technique that integrated $B^+-tree$ proposed by Bayer for indexing simple attributes and inverted file structure for text medical keywords acquired from additional description information about Radiological images. In our results, we implemented proposed retrieval system with Delphi under Windows XP environment. End users, Radiological technologists, are able to store simple attributes information such as doctor name, operator name, body parts, disease and so on, additional text-based description information, and Radiological image itself as well as to retrieve wanted results by using simple attributes and text keywords from large image databases by graphic user interface. Consequently proposed system can be used for effective clinical decision on Radiological image, reduction of education time by organizing the knowledge, and well organized education in the clinical fields. In addition, It can be expected to develop as decision support system by constructing web-based integrated imaging system included general image and special contrast image for the future.

• Journal of the Korean Society for information Management
• /
• v.23 no.3 s.61
• /
• pp.5-21
• /
• 2006

In recent years, theories of image and sound analysis have been proposed to work with text retrieval systems and have progressed quickly with the rapid progress in data processing speeds. This study proposes a common representation format for multimedia documents (CRFMD) composed of both images and text to form a single data structure. It also shows that image classification of a given test set is dramatically improved when text features are encoded together with image features. CRFMD might be applicable to other areas of multimedia document retrieval and processing, such as medical image retrieval, World Wide Web searching, and museum collection retrieval.

• KIISE Transactions on Computing Practices
• /
• v.22 no.6
• /
• pp.290-295
• /
• 2016

Recently, folksonomy-based image-sharing sites where users cooperatively make and utilize tags of image annotation have been gaining popularity. Typically, these sites retrieve images for a user request using simple text-based matching and display retrieved images in the form of photo stream. However, these tags are personal and subjective and images are not categorized, which results in poor retrieval accuracy and low user satisfaction. In this paper, we propose a categorization scheme for folksonomy images which can improve the retrieval accuracy in the tag-based image retrieval systems. Consequently, images are classified by the semantic similarity using text-information and image-information generated on the folksonomy. To evaluate the performance of our proposed scheme, we collect folksonomy images and categorize them using text features and image features. And then, we compare its retrieval accuracy with that of existing systems.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.451-454
• /
• 2003

A new query method, called query by visual keys, is proposed to aim easy operation and efficient region-based image retrieval (RBIR). Visual keys are constructed from representative regions/subimages in a given image database, and the database is indexed with visual keys. A system on PC is presented, where text retrieval techniques are applied to the image retrieval with visual keys. Experimental results show that one retrieval is done within 4ms and that the proposed system achieves the comparable retrieval precision (with user-friendly operation and low computational cost) to conventional region based image retrieval systems

• Proceedings of the KIEE Conference
• /
• 2006.10c
• /
• pp.521-523
• /
• 2006

This paper proposes an algorithm and its data structure to support real-time full-text search for the streamed or broadcasted multimedia data containing real-time stenograph text. Since the traditional indexing method used at information retrieval area uses the linguistic information, there is a heavy cost. Therefore, we propose the algorithm and its data structure based on suffix array, which is a simple data structure and has low space complexity. Suffix array is useful frequently to search for huge text. However, subtitle text of multimedia data is to get longer by time. Therefore, suffix array must be reconstructed because subtitle text is continually changed. We propose the data structure called prefix array and search algorithm using it.

• Journal of the Korean Society for information Management
• /
• v.27 no.1
• /
• pp.61-77
• /
• 2010

Blogs are now one of the major information resources on the web. The purpose of this study is to enhance the performance of blog retrieval by means of user assigned tags and trackback information. To this end, retrieval experiments were performed with a dataset of 4,908 blog pages together with their associated trackback URLs. In the experiments, text terms, user tags, and network centrality values based on trackbacks were variously combined as retrieval features. The experimental results showed that employing user tags and network centrality values as retrieval features in addition to text words could improve the performance of blog retrieval.

• The KIPS Transactions:PartB
• /
• v.11B no.5
• /
• pp.573-580
• /
• 2004

As the modern techniques have improved, people intend to store and manage the information on the web. Especially, it is the image data that is given a great deal of weight of the information because of the development of the scan and popularization of the digital camera and the cell-phone's camera. However, most image retrieval systems are still based on the text annotations while many images are creating everyday on the web. In this paper, we suggest the new approach for the semantic image retrieval using the RDF metadata based on the representation of the spatial relationships. For the semantic image retrieval, firstly we define the new vocabularies to represent the spatial relationships between the objects in the image. Secondly, we write the metadata about the image using RDF and new vocabularies. Finally. we could expect more correct result in our image retrieval system.