A5104831088- Data Mining Algorithms and Applications
- Advanced Image and Video Retrieval Techniques
- Rough Sets and Fuzzy Logic
- Image Retrieval and Classification Techniques
- Time Series Analysis and Forecasting
- Anomaly Detection Techniques and Applications
- Data Management and Algorithms
- Remote-Sensing Image Classification
- Advanced Database Systems and Queries
- Advanced Bandit Algorithms Research
- Machine Learning and Data Classification
- Data Stream Mining Techniques
- Natural Language Processing Techniques
- Advanced Neural Network Applications
- Imbalanced Data Classification Techniques
- Domain Adaptation and Few-Shot Learning
- Video Analysis and Summarization
- Neural Networks and Applications
- Stock Market Forecasting Methods
- Machine Learning and Algorithms
- Robotics and Sensor-Based Localization
- Video Surveillance and Tracking Methods
- Handwritten Text Recognition Techniques
- Advanced Vision and Imaging
- Algorithms and Data Compression
Institut de Recherche en Informatique et Systèmes Aléatoires
2004-2024
Université de Rennes
2018-2024
Institut national de recherche en informatique et en automatique
2018-2024
Centre National de la Recherche Scientifique
2012-2024
Institut Universitaire de France
2021-2023
Inserm
2023
Laboratoire Hubert Curien
2009-2022
Czech Academy of Sciences, Institute of Computer Science
2020
Université Claude Bernard Lyon 1
2009-2017
Lyon College
2017
Multispectral image pairs can provide complementary visual information, making pedestrian detection systems more robust and reliable. To benefit from both RGB thermal IR modalities, we introduce a novel attentive multispectral feature fusion approach. Under the guidance of inter- intra-modality attention modules, our deep learning architecture learns to dynamically weigh fuse features. Experiments on two public multi-spectral object datasets demonstrate that proposed approach significantly...
Multispectral images (e.g. visible and infrared) may be particularly useful when detecting objects with the same model in different environments day/night outdoor scenes). To effectively use spectra, main technical problem resides information fusion process. In this paper, we propose a new halfway feature method for neural networks that leverages complementary/consistency balance existing multispectral features by adding to network architecture, particular module cyclically fuses refines...
This paper presents GridNet, a new Convolutional Neural Network (CNN) architecture for semantic image segmentation (full scene labelling). Classical neural networks are implemented as one stream from the input to output with subsampling operators applied in order reduce feature maps size and increase receptive field final prediction. However, segmentation, where task consists providing class each pixel of an image, reduction is harmful because it leads resolution loss To tackle this problem,...
Multivariate Time Series (MTS) classification has gained importance over the past decade with increase in number of temporal datasets multiple domains. The current state-of-the-art MTS classifier is a heavyweight deep learning approach, which outperforms second-best only on large datasets. Moreover, this approach cannot provide faithful explanations as it relies post hoc model-agnostic explainability methods, could prevent its use numerous applications. In paper, we present XCM, an...
Bag-of-words-based image classification approaches mostly rely on low level local shape features. However, it has been shown that combining multiple cues such as color, texture, or is a challenging and promising task which can improve the accuracy. Most of state-of-the-art feature fusion methods usually aim to weight without considering their statistical dependence in application at hand. In this paper, we present new logistic regression-based method, called LRFF, takes advantage different...
We present DL8, an exact algorithm for finding a decision tree that optimizes ranking function under size, depth, accuracy and leaf constraints. Because the discovery of optimal trees has high theoretical complexity, until now few efforts have been made to compute such real-world datasets. An is both scientific practical interest. From point view, it can be used as gold standard evaluate performance heuristic constraint-based learners gain new insight in traditional learners. application...
Biological response modifiers (BRMs), secreted by platelets (PLTs) during storage, play a role in adverse events (AEs) associated with transfusion. Moreover, mitochondrial DNA (mtDNA) levels PLT components (PCs) are AEs. In this study we explore whether there is correlation between pathogenic BRMs and mtDNA these markers can be considered predictors of transfusion pathology.We investigated series reported AEs after PC transfusion, combining clinical observations mathematical modeling...
Platelet component (PC) transfusion leads occasionally to inflammatory hazards. Certain BRMs that are secreted by the platelets themselves during storage may have some responsibility.First, we identified non-stochastic arrangements of platelet-secreted in platelet components led acute reactions (ATRs). These data provide formal clinical evidence generate secretion profiles under both sterile activation and pathological conditions. We next aimed predict risk hazardous outcomes establishing...
Color constancy is the ability of human visual system to perceive constant colors for a surface despite changes in spectrum illumination. In computer vision, main approach consists estimating illuminant color and then remove its impact on objects. Many image processing algorithms have been proposed tackle this problem automatically. However, most these approaches are handcrafted mostly rely strong empirical assumptions, e.g., that average reflectance scene gray. State-of-the-art can perform...
We propose a novel approach to improve the reproducibility of neuroimaging results by converting statistic maps across different functional MRI pipelines. make assumption that pipelines can be considered as style component data and use generative models, among which, Diffusion Models (DM) convert between design new DM-based unsupervised multi-domain image-to-image transition framework constrain generation 3D fMRI using latent space an auxiliary classifier distinguishes from extend...
Finding an appropriate image representation is a crucial problem in robotics. This has been classically addressed by means of computer vision techniques, where local and global features are used. The selection or/and combination different carried out taking into account repeatability distinctiveness, but also the specific to solve. In this article, we propose generation descriptors from general purpose semantic annotations. approach evaluated as source information for scene classifier,...
Our research tackles the challenge of milk production resource use efficiency in dairy farms with machine learning methods. Reproduction is a key factor for farm performance since cows begin birth calf. Therefore, detecting estrus, only period when cow susceptible to pregnancy, crucial efficiency. goal enhance estrus detection (performance, interpretability), especially on currently undetected silent (35% total estrus), and allow farmers rely automatic solutions based affordable data...