Amsaleg Laurent - MultiMedia Modeling: 23rd International Conference, MMM 2017, Reykjavik, Iceland, January 4-6, 2017, Proceedings, Part II

MultiMedia Modeling: 23rd International Conference, MMM 2017, Reykjavik, Iceland, January 4-6, 2017, Proceedings, Part II — read online for free the complete book (whole text) full work

Below is the text of the book, divided by pages. System saving the place of the last page read, allows you to conveniently read the book "MultiMedia Modeling: 23rd International Conference, MMM 2017, Reykjavik, Iceland, January 4-6, 2017, Proceedings, Part II" online for free, without having to search again every time where you left off. Put a bookmark, and you can go to the page where you finished reading at any time.

Light

Font size:

↓

↑

Georgia Tahoma Arial Verdana

Reset

Interval:

↓

↑

Bookmark:

Make

1234567...176

Full Papers Accepted for Poster Presentation

Springer International Publishing AG 2017

Laurent Amsaleg , Gylfi r Gumundsson , Cathal Gurrin , Bjrn r Jnsson and Shinichi Satoh (eds.) MultiMedia Modeling Lecture Notes in Computer Science 10133 10.1007/978-3-319-51814-5_1

A Comparative Study for Known Item Visual Search Using Position Color Feature Signatures

Jakub Loko 1

(1)

SIRET Research Group, Department of Software Engineering, Faculty of Mathematics and Physics, Charles University in Prague, Prague, Czech Republic

Jakub Loko (Corresponding author)

Email:

David Kubo

Email:

Adam Blaek

Email:

Abstract

According to the results of the Video Browser Showdown competition, position-color feature signatures proved to be an effective model for visual known-item search tasks in BBC video collections. In this paper, we investigate details of the retrieval model based on feature signatures, given a state-of-the-art known item search tool Signature-based Video Browser. We also evaluate a preliminary comparative study for three variants of the utilizes distance measures. In the discussion, we analyze logs and provide clues for understanding the performance of our model.

Keywords

Similarity search Feature extraction Known item search Color sketch

Introduction

Nowadays, video data are present almost everywhere which challenges state-of-the-art video management systems and their query interfaces. Whereas traditional approaches rely on text-based query formulation or query by example paradigm []. An example of such scenario is known-item search (or mental query retrieval), where users cannot perfectly materialize their search intents and try to iteratively interact with the system to find a desired scene.

Systems for solving known-item search tasks rely on intuitive, interactive and multi-modal query interfaces. In the known-item search process, the user is in the center of the retrieval process and controls the intermediate actions navigating him towards the results. This feature complicates development and evaluation of known-item search systems. Therefore, competitions like the Video Browser Showdown []) are organized, where participating teams compete in predefined known-item search tasks. Two popular tasks are visual and textual known item search. In the visual known item search task, users see and memorize a short video clip (recording is not allowed), while in the textual known item search task, users receive a short text describing the desired scene.

During the last five years of the Video Browser Showdown, several promising approaches have been revealed by winning teams. Generally, the winning tools [] pointed on several features to be highly competitive in known-item search tasks:

• Effective query initialization instead of starting the search from the scratch, a preliminary initialization is necessary. For example, position-color sketches have proved to be highly effective for visual known-item search [].
• Concept based filtering restricting the collection to scenes containing recognized concepts significantly helps with the retrieval. However, the technique relies on the effectiveness of concept detectors and their ability to recognize a search concept. In this direction, novel deep learning based approaches achieve promising results [].
• Visualization of the results after query initialization and concept based filtering, the results require a suitable form of visualization enabling fast detection of desired scenes. A popular approaches are color sorted image maps [] and/or results accompanied with temporal context from the corresponding video.
• Effective browsing the correct results are often not present on the first page and so scrolling between pages becomes necessary. Furthermore, scrolling can be performed also within the temporal context of a detected results, using similarity search techniques given a candidate key frame, or other video interaction scenarios [].

Available work []. We summarize the tool in the next section and highlight the features investigated in this paper. Then we introduce three different distance measures that can be utilized in our model. To select the most appropriate one, we carried out a user study which is then analyzed and discussed. Alongside the distance measure selection, we provide a number of observations and clues on user behaviour leading to successful search.

Signature-Based Video Browser

The Signature-based Video Browser tool (SBVB) [ left). Hence, similar key frames from different parts of the video can be distinguished.

Fig. 1.

Sketch-based Video Browser user interface (left) and a detail of sketching canvas (right). (Color figure online)

Since 2014, the tool has been significantly extended by several new features [ left). No other visualization technique was used in the study.

Feature Signatures Video Retrieval Model

In this section, details of the retrieval model for a set of video files are presented.

To index video files, roughly one key frame for every second is selected, resulting in the set of key frames . For all the key frames , feature signatures are extracted where denotes the j -th centroid of the i -th feature signature. The centroid is defined as a tuple comprising x , y coordinates of the circle, L , a , b color coordinates from the CIE Lab color space and r denoting radius of the circle. The feature signatures are extracted using an adaptive k-means clustering. As mentioned earlier, users are enabled to define several sketch centroids that are matched to the extracted feature signatures. Since users may memorize only the most distinct color regions from the searched scene, only few query centroids are expected to be specified; hence, the model uses local instead of global matching between two feature signatures.

For a user defined query and a centroid distance measure , we start with calculating distances to all the key frames for each of the sketch centroids separately. For the sketch centroid the distance to the i -th key frame is defined as:

(1)

I.e.,

Next page

1234567...176

Light

Font size:

↓

↑

Georgia Tahoma Arial Verdana

Reset

Interval:

↓

↑

Bookmark:

Make

Similar books «MultiMedia Modeling: 23rd International Conference, MMM 2017, Reykjavik, Iceland, January 4-6, 2017, Proceedings, Part II»

Look at similar books to MultiMedia Modeling: 23rd International Conference, MMM 2017, Reykjavik, Iceland, January 4-6, 2017, Proceedings, Part II. We have selected literature similar in name and meaning in the hope of providing readers with more options to find new, interesting, not yet read works.

Bernd Finkbeiner

Verification, Model Checking, and Abstract Interpretation: 23rd International Conference, VMCAI 2022, Philadelphia, PA, USA, January 16–18, 2022, Proceedings

Jie Chen

Distributed Artificial Intelligence: Third International Conference, DAI 2021, Shanghai, China, December 17–18, 2021, Proceedings

Enver Ever (editor)

Forthcoming Networks and Sustainability in the IoT Era: First EAI International Conference, FoNeS – IoT 2020, Virtual Event, October 1-2, 2020, Proceedings

Nicolas Tsapatsoulis

Computer Analysis of Images and Patterns: 19th International Conference, CAIP 2021, Virtual Event, September 28–30, 2021, Proceedings, Part I

De-Shuang Huang

Intelligent Computing Theories and Application: 17th International Conference, ICIC 2021, Shenzhen, China, August 12–15, 2021, Proceedings, Part II

Bernabé Dorronsoro

Optimization and Learning: 4th International Conference, OLA 2021, Catania, Italy, June 21-23, 2021, Proceedings

Anthony Brooks

Interactivity and Game Creation: 9th EAI International Conference, ArtsIT 2020, Aalborg, Denmark, December 10–11, 2020, Proceedings

Ferruccio Damiani

Coordination Models and Languages: 23rd IFIP WG 6.1 International Conference, COORDINATION 2021, Held as Part of the 16th International Federated ...

Rafael Pass

Theory of Cryptography: 18th International Conference, TCC 2020, Durham, NC, USA, November 16–19, 2020, Proceedings, Part III

Korytkowski Marcin.

Artificial Intelligence and Soft Computing: 16th International Conference, ICAISC 2017, Zakopane, Poland, June 11-15, 2017, Proceedings, Part I

Grünbacher Paul

Requirements Engineering: Foundation for Software Quality: 23rd International Working Conference, REFSQ 2017, Essen, Germany, February 27 - March 2, 2017, Proceedings

Amann Johanna

Passive and Active Measurement: 18th International Conference, PAM 2017, Sydney, NSW, Australia, March 30-31, 2017, Proceedings

Discussion, reviews of the book MultiMedia Modeling: 23rd International Conference, MMM 2017, Reykjavik, Iceland, January 4-6, 2017, Proceedings, Part II and just readers' own opinions. Leave your comments, write what you think about the work, its meaning or the main characters. Specify what exactly you liked and what you didn't like, and why you think so.