Long-Term Incremental Web-Supervised Learning of Visual Concepts via Random Savannas

Ewerth, Ralph; Ballafkir, K.; Mühling, Markus; Seiler, Dominik; Freisleben, Bernd

doi:10.1109/tmm.2012.2186956

Cited by 10 publications

(4 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hashing, surveyed by Zhang and Rui [98], is often used in order to reduce dimensionality. The second step of the pipeline, classification, is in almost all cases carried out by a support vector machine (SVM) by Cortes and Vapnik [14], although other classifiers such as nearest neighbour [6] or random forests [22] have also been used. The second pipeline, deep learning, revolves around using deep neural nets, which extract features and semantics using one model.…”

Section: Representation and Learningmentioning

confidence: 99%

Towards interactive, intelligent, and integrated multimedia analytics

Zahálka

Worring

2014

2014 IEEE Conference on Visual Analytics Science and Technology (VAST)

View full text Add to dashboard Cite

Abstract-The size and importance of visual multimedia collections grew rapidly over the last years, creating a need for sophisticated multimedia analytics systems enabling large-scale, interactive, and insightful analysis. These systems need to integrate the human's natural expertise in analyzing multimedia with the machine's ability to process large-scale data. The paper starts off with a comprehensive overview of representation, learning, and interaction techniques from both the human's and the machine's point of view. To this end, hundreds of references from the related disciplines (visual analytics, information visualization, computer vision, multimedia information retrieval) have been surveyed. Based on the survey, a novel general multimedia analytics model is synthesized. In the model, the need for semantic navigation of the collection is emphasized and multimedia analytics tasks are placed on the explorationsearch axis. The axis is composed of both exploration and search in a certain proportion which changes as the analyst progresses towards insight. Categorization is proposed as a suitable umbrella task realizing the exploration-search axis in the model. Finally, the pragmatic gap, defined as the difference between the tight machine categorization model and the flexible human categorization model is identified as a crucial multimedia analytics topic.Index Terms-Multimedia (image/video/music) visualization, machine learning.

show abstract

Section: Representation and Learningmentioning

confidence: 99%

Towards interactive, intelligent, and integrated multimedia analytics

Zahálka

Worring

2014

2014 IEEE Conference on Visual Analytics Science and Technology (VAST)

View full text Add to dashboard Cite

show abstract

“…Jasper et al [4] proposed a real time visual concept for classification. Ewerth et al [11] proposed long term incremental web supervised learning of visual concepts via random savannas. Chen et al [12] proposed automatic training image acquisition and effective feature selection from community contributed photos for facial attribute detection.…”

Section: Related Workmentioning

confidence: 99%

Rough-fuzzy based scene categorization for text detection and recognition in video

Roy

Shivakumara

Jain

et al. 2018

Pattern Recognition

View full text Add to dashboard Cite

Scene image or video understanding is a challenging task especially when number of video types increases drastically with high variations in background and foreground. This paper proposes a new method for categorizing scene videos into different classes, namely, Animation, Outlet, Sports, e-Learning, Medical, Weather, Defense, Economics, Animal Planet and Technology, for the performance improvement of text detection and recognition, which is an effective approach for scene image or video understanding. For this purpose, at first, we present a new combination of rough and fuzzy concept to study irregular shapes of edge components in input scene videos, which helps to classify edge components into several groups. Next, the proposed method explores gradient direction information of each pixel in each edge component group to extract stroke based features by dividing each group into several intra and inter planes. We further extract correlation and covariance features to encode semantic features located inside planes or between planes. Features of intra and inter planes of groups are then concatenated to get a feature matrix. Finally, the feature matrix is verified with temporal frames and fed to a neural network for categorization. Experimental results show that the proposed method outperforms the existing state-of-the-art methods, at the same time, the performances of text detection and recognition methods are also improved significantly due to categorization.

show abstract

“…An image is filtered out if one of the two filters (textual or visual) considers an image as spam. Spam detection based on textual information is described in previous work [7]. The crawling of web content and the spam detection steps are conducted separately for each specified target concept.…”

Section: Web-supervised Learningmentioning

confidence: 99%

On the Effects of Spam Filtering and Incremental Learning for Web-Supervised Visual Concept Classification

Springstein

Ewerth

2016

Proceedings of the 2016 ACM on International Conference on Multimedia Retrieval

Self Cite

View full text Add to dashboard Cite

Deep neural networks have been successfully applied to the task of visual concept classification. However, they require a large number of training examples for learning. Although pre-trained deep neural networks are available for some domains, they usually have to be fine-tuned for an envisaged target domain. Recently, some approaches have been suggested that are aimed at incrementally (or even endlessly) learning visual concepts based on Web data. Since tags of Web images are often noisy, normally some filtering mechanisms are employed in order to remove "spam" images that are not appropriate for training. In this paper, we investigate several aspects of a web-supervised system that has to be adapted to another target domain: 1.) the effect of incremental learning, 2.) the effect of spam filtering, and 3.) the behavior of particular concept classes with respect to 1.) and 2.). The experimental results provide some insights under which conditions incremental learning and spam filtering are useful.

show abstract

Long-Term Incremental Web-Supervised Learning of Visual Concepts via Random Savannas

Cited by 10 publications

References 25 publications

Towards interactive, intelligent, and integrated multimedia analytics

Towards interactive, intelligent, and integrated multimedia analytics

Rough-fuzzy based scene categorization for text detection and recognition in video

On the Effects of Spam Filtering and Incremental Learning for Web-Supervised Visual Concept Classification

Contact Info

Product

Resources

About