A5081523913- Human Pose and Action Recognition
- Context-Aware Activity Recognition Systems
- Video Surveillance and Tracking Methods
- Anomaly Detection Techniques and Applications
- Advanced Image and Video Retrieval Techniques
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Plant Pathogens and Fungal Diseases
- Dementia and Cognitive Impairment Research
- Human-Automation Interaction and Safety
- Spectroscopy Techniques in Biomedical and Chemical Research
- Smart Agriculture and AI
- Transportation and Mobility Innovations
- Autonomous Vehicle Technology and Safety
- Infrared Thermography in Medicine
- Remote-Sensing Image Classification
- Zoonotic diseases and public health
- Domain Adaptation and Few-Shot Learning
- Multimodal Machine Learning Applications
- Plant Physiology and Cultivation Studies
- Image and Video Stabilization
- Gait Recognition and Analysis
- Stroke Rehabilitation and Recovery
- French Language Learning Methods
- 3D Surveying and Cultural Heritage
Laboratoire d'Informatique en Images et Systèmes d'Information
2019-2024
Centre National de la Recherche Scientifique
2019-2024
Institut National des Sciences Appliquées de Lyon
2022-2024
Université Claude Bernard Lyon 1
2019-2024
Université Lumière Lyon 2
2018-2021
Institut national de recherche en informatique et en automatique
2012-2020
Université Côte d'Azur
2017
Cognition Behaviour Technology
2015-2017
KU Leuven
2017
Research Centre Inria Sophia Antipolis - Méditerranée
2015
The use of Serious Games (SG) in the health domain is expanding. In field neurodegenerative disorders (ND) such as Alzheimer's disease, SG are currently employed both to support and improve assessment different functional cognitive abilities, provide alternative solutions for patients' treatment, stimulation, rehabilitation. As quite young, recommendations on people with ND still rare. 2014 we proposed some initial (Robert et al., 2014). aim present work was update them, thanks opinions...
The land cover reconstruction from monochromatic historical aerial images is a challenging task that has recently attracted an increasing interest the scientific community with proliferation of large-scale epidemiological studies involving retrospective analysis spatial patterns. However, efforts made by computer vision in remote-sensing applications are mostly focused on prospective approaches through high-resolution multi-spectral data acquired advanced programs. Hence, four contributions...
Combining multimodal concept streams from heterogeneous sensors is a problem superficially explored for activity recognition. Most studies explore simple in nearly perfect conditions, where temporal synchronization guaranteed. Sophisticated fusion schemes adopt problem-specific graphical representations of events that are generally deeply linked with their training data and focused on single sensor. This paper proposes hybrid framework between knowledge-driven probabilistic-driven methods...
Currently, the assessment of autonomy and functional ability involves clinical rating scales. However, scales are often limited in their to provide objective sensitive information. By contrast, information communication technologies may overcome these limitations by capturing more fully as well cognitive disturbances associated with Alzheimer disease (AD). We investigated quantitative dementia patients based not only on gait analysis but also participant performance instrumental activities...
The Computed Tomography Imaging Spectrometer (CTIS) permits a snapshot acquisition of hyperspectral cube, through the creation an image indirect measurements which is then traditionally used for reconstruction cube. This step time-consuming and only yields approximation original Following compressed learning framework, we compare performance classification task carried out on reconstructed cubes one hand, directly raw images other. Regarding latter case, propose in particular use new...
The task of 2D human pose estimation has known a significant gain performance with the advent deep learning. This aims to estimate body keypoints people in an image or video. However, real-life applications such methods bring new challenges that are under-represented general context datasets. For instance, driver status monitoring on consumer road vehicles introduces difficulties, like self- and background body-part occlusions, varying illumination conditions, cramped view angles, etc. These...
come from teaching and research institutions in France or abroad, public private centers.L'archive ouverte pluridisciplinaire
Visual activity recognition plays a fundamental role in several research fields as way to extract semantic meaning of images and videos. Prior work has mostly focused on classification tasks, where label is given for video clip. However, real life scenarios require method browse continuous flow, automatically identify relevant temporal segments classify them accordingly target activities. This paper proposes knowledge-driven event framework address this problem. The novelty the lies...
Industries nowadays have an increasing need of real-time and accurate vision-based algorithms.Although the performance object detection methods improved a lot thanks to massive public datasets, instance in industrial context must be approached differently, since annotated images are usually unavailable or rare.In addition, when video stream comes from head-mounted camera, we observe movements blurred frames altering image content.For this purpose, propose framework generate dataset...
Background: At present, the assessment of autonomy in daily living activities, one key symptoms Alzheimer’s disease (AD), involves clinical rating scales. Methods: In total, 109 participants were included. particular, 11 during a pre-test Nice, France, and 98 (27 AD, 38 mild cognitive impairment—MCI—and 33 healthy controls—HC) Thessaloniki, Greece, carried out standardized scenario consisting several instrumental activities (IADLs), such as making phone call or preparing pillbox while being...
The computed tomography imaging spectrometer (CTIS) is a snapshot hyperspectral system. Its output 2D image of multiplexed spatiospectral projections the cube scene. Traditionally, 3D reconstructed from this before further analysis. In paper, we show that it possible to learn information directly CTIS raw output, by training neural network perform binary classification on such images. use case study an agricultural one, as imagery used substantially in field: detection apple scab lesions...
We herein present a hierarchical model-based framework for event recognition using multiple sensors. Event models combine priori knowledge of the scene (3D geometric and semantic information, such as contextual zones equipment) with moving objects (e.g., Person) detected by monitoring system. The follow generic ontology based on natural language, which allows domain experts to easily adapt them. novelty relies combining sensors at decision (event) level, handling their conflict probabilistic...
The interest in driver monitoring has grown recently, especially the context of autonomous vehicles.However, training deep neural networks for computer vision requires more and images with significant diversity, which does not match reality field.This lack data prevents to be properly trained certain complex tasks such as human pose transfer aims produce an image a person target from another same person.To tackle this problem, we propose new synthetic dataset pose-related tasks.By using...
The adoption of self-driving cars (SDC) will certainly revolutionize our lives, even though they may take more time to become fully autonomous than initially predicted. first vehicles are already present in certain cities the world, as part experimental robot-taxi services. However, most existing studies focus on navigation such vehicles. We currently miss methods, datasets, and assess in-cabin human component technology real-world conditions. This paper proposes an framework study...