A5014303032- Advanced Image and Video Retrieval Techniques
- Video Analysis and Summarization
- Multimodal Machine Learning Applications
- Image Retrieval and Classification Techniques
- Artificial Intelligence in Games
- Digital Games and Media
- Human Pose and Action Recognition
- AI-based Problem Solving and Planning
- Sports Analytics and Performance
- 3D Surveying and Cultural Heritage
- 3D Shape Modeling and Analysis
- Domain Adaptation and Few-Shot Learning
- Constraint Satisfaction and Optimization
- Computer Graphics and Visualization Techniques
- Recommender Systems and Techniques
- Data Visualization and Analytics
- Educational Games and Gamification
- Algorithms and Data Compression
- Data Management and Algorithms
- Advanced Graph Neural Networks
University of Basel
2018-2024
ETH Zurich
1991-1996
Despite the fact that automatic content analysis has made remarkable progress over last decade - mainly due to significant advances in machine learning interactive video retrieval is still a very challenging problem, with an increasing relevance practical applications. The Video Browser Showdown (VBS) annual evaluation competition pushes limits of state-of-the-art tools, tasks, data, and metrics. In this paper, we analyse results outcome 8th iteration VBS detail. We first give overview novel...
This work summarizes the findings of 7th iteration Video Browser Showdown (VBS) competition organized as a workshop at 24th International Conference on Multimedia Modeling in Bangkok. The focuses video retrieval scenarios which searched scenes were either previously observed or described by another person (i.e., an example shot is not available). During event, nine teams competed with their tools providing access to shared collection 600 hours content. Evaluation objectives, rules, scoring,...
Multimedia retrieval and analysis are two important areas in "Big data" research. They have common that they work with feature vectors as proxies for the media objects themselves. Together metadata such textual descriptions or numbers, these describe a object its entirety, must therefore be considered jointly both storage retrieval.
The steady growth of multimedia collections - both in terms size and heterogeneity necessitates systems that are able to conjointly deal with several types media as well large volumes data. This is especially true when it comes satisfying a particular information need, i.e., retrieving object interest from collection. Nevertheless, existing management retrieval mostly organized silos treat different separately. Hence, they limited crossing these for accessing objects. In this paper, we...
With the increase in sensory capability of mobile devices, data that can be generated and used a lifelogging context gets increasingly diverse. Such is special multimedia, not only because its close personal relationship with originator, but also diverse multimodality composition from structured, semi-structured, unstructured data. This diversity poses retrieval challenges are unique to lifelog which have implications for activity other multimedia domains. In this paper, we present...
The variety and amount of data being collected in our everyday life poses unique challenges for multimedia retrieval. In the Lifelog Search Challenge (LSC), retrieval systems compete finding events based on descriptions containing hints about structured, semi-structured an unstructured data. this paper, we present system vitrivr with a focus changes additions made new dataset, successful participation at LSC 2019. Specifically, show how dataset can be used different modalities without...
The Lifelog Search Challenge (LSC) is an annual benchmarking competition for interactive multimedia retrieval systems, where participating systems compete in finding events based on textual descriptions containing hints about structured, semi-structured, and/or unstructured data. In this paper, we present the system vitrivr, a long-time participant to LSC, with focus new functionality. Specifically, introduce image stabilisation module which added prior feature extraction reduce degradation...
The game of Nine Men's Morris is a draw. We obtained this result using combination endgame databases (10 10 states) and search. Our improved algorithm for computing allowed the to be solved on personal computer. Other games have been knowledge‐based methods dramatically prune search tree. does not seem profit from such methods, making it first nontrivial in which almost entire state space has considered.
The multimodal nature of lifelog data collections poses unique challenges for multimedia management and retrieval systems. Lifelog Search Challenge (LSC) offers an annual evaluation platform such interactive They compete against one another in finding items interest within a set time frame.
Personal lifelog data collections are becoming more common as a memory aid, well for analytical tasks, such health and fitness analysis. Due to the multimodal personal nature of data, interactive multimedia retrieval approaches required facilitate flexible iterative query formulation result exploration In recent years, novel user interface modalities have emerged, that allow new ways users interact with system. Virtual reality, one modality, provides advantages challenges in comparison...
The collection of lifelog data --- visual and multi-sensory data, including biometric spatiotemporal metadata becomes easier more supported by commercial products every year. Naturally, is multi-modal, with arguably a major audio-visual component, such as captured videos, audio recordings photos. For retrieval, the challenges managing accessing (visual) multimedia content are paired semi-structured heterogeneous metadata. One approach to these application general-purpose, content-based...
The digitization of museum exhibits has raised the question how to make these data accessible, particularly in light ever growing collections being available. In this demo, we present VIRTUE system which allows curators easily set up virtual exhibitions static and dynamic 2D (paintings, photographs, videos, etc.) 3D artifacts. Visitors may navigate through rooms, inspect artifacts interact with them novel ways. Participants will be able use by creating their own exhibitions, they tour as a visitor.
The multi-modal and interrelated nature of lifelog data makes it well suited for graph-based representations. In this paper, we present the second iteration LifeGraph, a Knowledge Graph Lifelog Data, initially introduced during 3rd Search Challenge in 2020. This incorporates several lessons learned from previous version. While actual graph has undergone only small changes, mechanisms by which is traversed querying as underlying storage system performs traversal have been changed. means query...
In the research of video retrieval systems, comparative assessments during dedicated competitions provide priceless insights into performance individual systems. The scope and depth such evaluations are unfortunately hard to improve, due limitations by set-up costs, logistics, organization complexity large events. We show that this easily impairs statistical significance collected results, reproducibility competition outcomes. article, we present a methodology for remote content-based...
Interactive retrieval with user-friendly and performant interfaces remains a necessity for video retrieval, even in light of significant gains performance through multi-modal encoders. In recent years, novel interaction modalities such as virtual reality (VR) augmented (AR) have gained popularity, but the best way to adapt paradigms from traditional interfaces, especially result browsing interaction, an open research question. this paper, we compare two controlled setting gain insight into...