A5068128546- Video Analysis and Summarization
- Image and Video Quality Assessment
- Human Pose and Action Recognition
- Sports Analytics and Performance
- Network Traffic and Congestion Control
- Natural Language Processing Techniques
- Wireless Networks and Protocols
- Advanced MIMO Systems Optimization
- Multimedia Communication and Technology
- Mobile Crowdsensing and Crowdsourcing
- Music and Audio Processing
- Anomaly Detection Techniques and Applications
- Power Line Communications and Noise
- Sports Performance and Training
- Video Surveillance and Tracking Methods
- Caching and Content Delivery
- Sports injuries and prevention
- Adversarial Robustness in Machine Learning
- Digital Games and Media
- Digital Media Forensic Detection
- Visual Attention and Saliency Detection
- Peer-to-Peer Network Technologies
- Green IT and Sustainability
- Video Coding and Compression Technologies
- Methane Hydrates and Related Phenomena
ForzaSys (Norway)
2023-2025
Simula Metropolitan Center for Digital Engineering
2021-2024
OsloMet – Oslo Metropolitan University
2024
Simula Research Laboratory
2017-2021
TU Wien
2015-2016
Social media plays a significant role for sports organizations with millions of active fans, but publishing highlights is often tedious manual operation. With the development AI, new tools are available content generation and personalization to engage audiences. We propose an AI-based multimedia production framework automatic soccer ice hockey on social media, disseminate our experiences developing dedicated pipelines event detection classification, player tracking, highlight clipping,...
Mobile Broadband (MBB) Networks are evolving at a fast pace, with technology enhancements that promise drastic improvements in capacity, connectivity, coverage, i.e., better performance general. But how to measure the actual of MBB solution? In this paper, we present our experience running simplest test: "speedtest-like" measurements estimate download speed offered by 3G/4G networks. Despite their simplicity, networks much more complex than wired networks, because additional factors (e.g.,...
Sports multimedia plays a central role in captivating audiences on social media platforms. However, fast-paced sports such as ice hockey pose unique challenges due to their swift gameplay and the small puck size, making object tracking-based video adaptation for complex task. In this context, we introduce SmartCrop-H, an innovative cropping tool powered by advanced AI models. It excels at tracking ensuring that crucial remains center of attention, regardless desired target aspect ratio. The...
Abstract Data analysis for athletic performance optimization and injury prevention is of tremendous interest to sports teams the scientific community. However, data are often sparse hard obtain due legal restrictions, unwillingness share, lack personnel resources be assigned tedious process curation. These constraints make it difficult develop automated systems analysis, which require large datasets learning. We therefore present SoccerMon , largest soccer athlete dataset available today...
Detecting events in videos is a complex task, and many different approaches, aimed at large variety of use-cases, have been proposed the literature. Most however, are unimodal only consider visual information videos. This paper presents evaluates approaches based on neural networks where we combine features with audio to detect (spot) classify soccer We employ model fusion modalities such as video audio, test these combinations against state-of-the-art models SoccerNet dataset. The results...
The rapid advancement of technology has been revolutionizing the field sports media, where there is a growing need for sophisticated data processing methods. Current methodologies extracting information from soccer broadcast videos to generate game highlights and summaries social media are predominantly manual rely heavily on text-based NLP techniques, overlooking rich visual auditory available. In response this challenge, our research introduces SoccerSum, tool that innovates in by...
Soccer is one of the most popular sports globally, and amount soccer-related content worldwide, including video footage, audio commentary, team/player statistics, scores, rankings, enormous rapidly growing. Consequently, generation multimodal summaries tremendous interest for broadcasters fans alike, as a large percentage audiences prefer to follow only main highlights game. However, annotating important events producing often requires expensive equipment lot tedious, cumbersome, manual...
This paper introduces SoccerSum, a novel dataset aimed at enhancing object detection and segmentation in video frames depicting the soccer pitch, using footage from Norwegian Eliteserien league across 2021-2023. With goal of detecting elements beyond common entities existing datasets, such as ball, players referees, this includes additional annotations for net, corner flag posts, penalty mark. SoccerSum also key pitch areas boxes same frame sequences. Comprising 750 annotated with 10 classes...
The application of Automatic Speech Recognition (ASR) technology in soccer offers numerous opportunities for sports analytics. Specifically, extracting audio commentaries with ASR provides valuable insights into the events game, and opens door to several downstream applications such as automatic highlight generation. This paper presents SoccerNet-Echoes, an augmentation SoccerNet dataset automatically generated transcriptions from game broadcasts, enhancing video content rich layers textual...
As the demand for mobile connectivity continues to grow, need evaluate performance of Mobile Broadband (MBB) networks becomes stronger. In last years, "speed", quantified most commonly by data rate, gained popularity as widely accepted metric describe MBB performance. However, there is a lack consensus on how speed should be measured. this paper, we design and implement MONROE-Nettest dissect measurements, investigate effect different factors measurements in complex ecosystem. built an...
Summary Video streaming applications constitute a significant portion of the Internet traffic today, with mobile accounting for more than half online video views. The high share in current mix has prompted many studies that examine through measurements. However, performance depends on different factors at layers TCP/IP stack. For example, browser selection application layer or choice protocol transport can have impact performance. Furthermore, heavily underlying network conditions (eg, and...
Crowdsourcing is a novel paradigm which has applications in plethora of disciplines including software development, public administration, and communication technologies. In the field telecommunications, crowdsourcing makes viable addition to state art benchmarking methods for mobile networks, when combined with sensing approaches employ smartphones as end nodes. this paper, we review opportunities challenges using crowdsourced measurements from demonstrate some these generic aspects an open...
Professional sports achievements combine not only the individual physical abilities of athletes but also many modern technologies in areas such as medicine, equipment production, nutrition, and mental health monitoring. In this work, we address problem predicting soccer players' ability to perform, from subjective self-reported wellness parameters collected using a commercially deployed digital monitoring system called PmSys. We use 2 years data two Norwegian female teams, where players have...
COVID-19 has changed our lives and continues to have a drastic impact. It forces us impose restrictive social distancing measures leads rely more on the Internet. Mobile networks become one of major means access Internet for most us. We investigate performance 9 mobile in Europe around time when restrictions were imposed, considering web QoE, throughput, RTT, signal coverage. Our results show countries' are impacted by (as high as 46 % increase page load time) at different times based...
Developing systems for the automatic detection of events in video is a task which has gained attention many areas including sports. More specifically, event soccer videos been studied widely literature. However, there are still number shortcomings state-of-the-art such as high latency, making it challenging to operate at live edge. In this paper, we present an algorithm detect real time, using 3D convolutional neural networks. We test our on three different datasets from SoccerNet, Swedish...
In the rapidly evolving landscape of digital platforms, need for optimizing media representations to cater various aspect ratios is palpable. this paper, we pioneer an approach that utilizes object detection, scene outlier and interpolation smart cropping. Using soccer as a case study, our primary goal capture frame salience using (player ball) detection tracking AI models. To improve tracking, rely on understanding explore techniques. Our pipeline, called SmartCrop, efficient, supports...
Deep learning has achieved immense success in computer vision and the potential to help physicians analyze visual content for disease other abnormalities. However, current state of deep is very much a black box, making medical professionals skeptical about integrating these methods into clinical practice. Several have been proposed shed some light on boxes, but there no consensus opinion doctors that will consume explanations. This paper presents study asking their state-of-the-art...
Figure 1: Overview of the sources, curation, and applications Kvasir-VQA dataset.