A5062563309- Multimodal Machine Learning Applications
- Image Retrieval and Classification Techniques
- Advanced Image and Video Retrieval Techniques
- Anomaly Detection Techniques and Applications
- Face recognition and analysis
- Video Surveillance and Tracking Methods
- Human Pose and Action Recognition
- Perfectionism, Procrastination, Anxiety Studies
- Work-Family Balance Challenges
- Advanced Neural Network Applications
- Handwritten Text Recognition Techniques
- Complex Systems and Time Series Analysis
- Tuberculosis Research and Epidemiology
- Visual Attention and Saliency Detection
- Cell Image Analysis Techniques
- Stock Market Forecasting Methods
- Face Recognition and Perception
- Biomedical Text Mining and Ontologies
- Time Series Analysis and Forecasting
- Speech and Audio Processing
- Network Security and Intrusion Detection
- Generative Adversarial Networks and Image Synthesis
- Music and Audio Processing
- Image and Video Quality Assessment
- Evolutionary Psychology and Human Behavior
Universitatea Națională de Știință și Tehnologie Politehnica București
2017-2024
Wellcome Trust
2021
Dublin City University
2020
In this article, we report on the creation of a publicly available, common evaluation framework for Violent Scenes Detection (VSD) in Hollywood and YouTube videos. We propose robust data set, VSD96, with more than 96 hours video various genres, annotations at different levels detail (e.g., shot-level, segment-level), mid-level concepts blood, fire), pre-computed multi-modal descriptors, over 230 system output results as baselines. This is most comprehensive set available to date tailored VSD...
Financial markets have always been a point of interest for automated systems. Due to their complex nature, financial algorithms and fintech frameworks require vast amounts data accurately respond market fluctuations. This availability is tied the daily evolution, so it impossible accelerate its acquisition. In this article, we discuss several solutions augmenting datasets via synthesizing realistic time-series with help generative models. problem complex, since time series present very...
The advent of generative networks and their adoption in numerous domains communities have led to a wave innovation breakthroughs artificial intelligence machine learning. Generative Adversarial Networks (GANs) expanded the scope what is possible with learning, allowing for new applications areas such as computer vision, natural language processing, creative AI. GANs, particular, been used wide range tasks, including image video generation, data augmentation, style transfer, anomaly...
CCTV systems bring numerous advantages to security systems, but they require notable efforts from human operators in case of alarming events order detect the precise triggering moments. This paper proposes a system that can automatically trigger alarms when it detects abandoned luggage, person left baggage and then tracks suspicious throughout perimeter covered by system. The is based on Mask R-CNN has been tested with several backbone configurations. Wee valuate each subsystem independently...
Face verification aims to distinguish between genuine and imposter pairs of faces, which include the same or different identities, respectively. The performance reported in recent years gives impression that task is practically solved. Here, we revisit problem argue existing evaluation datasets were built using two oversimplifying design choices. First, usual identity selection form not challenging enough because, practice, needed detect imposters. Second, underlying demographics are often...
Automated lip-reading, i.e., translating lip movements into text, has received growing interest in recent years with the success of deep learning across a wide variety tasks. One major obstacle to progress this field been lack suitable training resources, vast majority being limited selective set languages. In paper, we study effectiveness transfer address massive amounts labeled data for building language-independent lip-reading system. Towards target, exploit existing knowledge and...
Training network models to accurately respond market fluctuations requires access vast amounts of data. Data availability is strictly bound the market's evolution, which updates only on a daily basis. In this paper, we propose several solutions based Generative Adversarial Networks for providing artificially generated time series data with realistic properties. The main challenge here specificity target data, has properties that are difficult control and have wide variations in time, e.g.,...
Embedded systems are under continuous development of innovative technological trends, such as adoption smart devices which becoming capable running complex video analytics tasks locally. Lately, deep neural networks have been successfully applied in the field computer vision achieving state-of-the-art results. These techniques not yet suitable for resource limited deployments due to high memory footprint and computational cost, factors that affect inference time. To tackle these constraints,...
Recent advancements in Generative Adversarial Networks (GANs) have enabled photorealistic image generation with high quality. However, the malicious use of such generated media has raised concerns regarding visual misinformation. Although deepfake detection research demonstrated accuracy, it is vulnerable to advances techniques and adversarial iterations on countermeasures. To address this, we propose a proactive sustainable training augmentation solution that introduces artificial...
Deep neural networks (DNNs) are universal estimators that have achieved state-of-the-art performance in a broad spectrum of classification tasks, opening new perspectives for many applications. One them is addressing ensemble learning. In this paper, we introduce set deep learning techniques with dense, attention, and convolutional network layers. Our approach automatically discovers patterns correlations between the decisions individual classifiers, therefore, alleviating difficulty...
Embedded systems are under continuous development of innovative technological trends, such as adoption smart devices which becoming capable running complex video analytics tasks locally. Lately, deep neural networks have been successfully applied in the field computer vision achieving state-of-the-art results. These techniques not yet suitable for resource limited deployments due to high memory footprint and computational cost, factors that affect inference time. To tackle these constraints,...
In this work, we consider the problem of person search, which is a challenging task that requires both detection and re-identification r un c oncurrently. context, propose search approach based on deep neural networks incorporates attention mechanisms to perform retrieval more robustly. Global local features are extracted for identification, respectively, boosted by layers allow extraction discriminative feature representations, all in an end-to-end manner. We evaluate our three data sets...