A5081068287- Multimodal Machine Learning Applications
- Advanced Image and Video Retrieval Techniques
- Human Pose and Action Recognition
- Hand Gesture Recognition Systems
- Video Analysis and Summarization
- Business Process Modeling and Analysis
- Information Technology Governance and Strategy
- Technology Adoption and User Behaviour
- Knowledge Management and Sharing
- Innovation and Knowledge Management
- Customer Service Quality and Loyalty
- Cancer-related molecular mechanisms research
- Image Retrieval and Classification Techniques
Sejong University
2005-2024
Korea Advanced Institute of Science and Technology
2005
In content-based video retrieval (CBVR), dealing with large-scale collections, efficiency is as important accuracy; thus, several video-level feature-based studies have actively been conducted. Nevertheless, owing to the severe difficulty of embedding a lengthy and untrimmed into single feature, these insufficient for accurate compared frame-level studies. this paper, we show that appropriate suppression irrelevant frames can provide insight current obstacles approaches. Furthermore, propose...
With the growth of video streaming industry, retrieval and alignment are facing high levels demand. Several studies have demonstrated feasibility these methods for various problems related to independently, but test in a unified framework has never been done. However, real-world applications, it is also concurrently necessary not only find which pairs similar (video retrieval) align positions pair that alignment). In this paper, we present new task simultaneously retrieves aligns videos. As...
As the demand for large-scale video analysis increases, retrieval research is also becoming more active. In 2014, ISO/IEC MPEG began standardizing compact descriptors analysis, known as CDVA, and it now adopted a standard. However, standardized CDVA not easily compared to other methods because MPEG-CDVA dataset used performance verification disclosed, despite fact that follow-up studies are underway with multiple versions of experimental model. addition, analyses modules constituting...
In content-based video retrieval (CBVR), dealing with large-scale collections, efficiency is as important accuracy; thus, several video-level feature-based studies have actively been conducted. Nevertheless, owing to the severe difficulty of embedding a lengthy and untrimmed into single feature, these insufficient for accurate compared frame-level studies. this paper, we show that appropriate suppression irrelevant frames can provide insight current obstacles approaches. Furthermore, propose...
Weakly supervised temporal action localization (WTAL) aims to detect instances in untrimmed videos using only video-level annotations. Since many existing works optimize WTAL models based on classification labels, they encounter the task discrepancy problem (i.e., localization-by-classification). To tackle this issue, recent studies have attempted utilize category names as auxiliary semantic knowledge through vision-language pre-training (VLP). However, there are still areas where research...