Fourth Workshop on "Robust Subspace Learning and Computer Vision", RSL-CV 2021 in conjunction with ICCV 2021, Montreal, Canada, October 2021. (More information)
First Workshop on "When Graph Signal Processing meets Computer Vision", GSP-CV 2021 in conjunction with ICCV 2021, Montreal, Canada, October 2021. (More information)
Special issue on "Neural Computing for IoT based Intelligent Healthcare Systems", Neural Computing and Application, to appear in 2021. (More information)
My research interests in computer vision consists mainly in the detection of moving objects in challenging surface (ground and sea), air and space environments. Sensors used are either passive sensors (i.e cameras) or active sensors (i.e radars) providing data images. More specifically, passive sensors are cameras in visible spectrum, IR cameras, multi-spectral cameras, hyper-spectral cameras, light-field cameras and event-based cameras. Active sensors are SAR, PolSAR, and LiDAR.
For detection of moving foreground objects in surface and air environments, my research focus on background subtraction (Background Subtraction Research). In this research, I investigated particularly the application of different mathematical concepts (statistical, fuzzy and Dempster-Schafer theories), machine learning concepts (reconstructive and discriminative subspace learning models, robust PCA and deep neural networks), and signal processing concepts (Graph Signal Processing) in the field of video surveillance.
Statistical, fuzzy and Dempster-Schafer theories allow to deal with imprecision, uncertainty and incompletness in the data images due the challenges (dynamic backgrounds, illumination changes to cite a few). Representation learning allows to deal with pertubations in the data images. Deep neural networks allow to have scene-specific or agnostic learning of scenes. Graph signal processing allows to reduce the required labeled data compared to deep neural networks.
My work also concerns full exhaustive surveys on mathematical and machine learning tools used in foreground/background separation. Furthermore, I investigated the field of decomposition into low-rank and additives matrices for background/foreground separation (DLAM Research), the field of decomposition into low-rank and additives tensors for background/foreground separation (DLAT research), and the field of decomposition into sparse and additive matrices for background/foreground separation (DSAM research).
My research concerns also robust texture features and feature selection for background/foreground separation (Features Research).
Keywords (Theory): Crisp concepts, Statistical concepts, Fuzzy concepts, Dempster-Schafer Theory, Robust Principal Component Analysis, Deep Neural Networks, Graph Signal Processing
M. Chapel, T. Bouwmans, "Moving Objects Detection with a Moving Camera: A Comprehensive Review", Computer Science Review, Volume 38, November 2020.
T. Bouwmans, S. Javed, M. Sultana, S. Jung, “Deep Neural Network Concepts in Background Subtraction: A Systematic Review and A Comparative Evaluation”, Neural Networks, 2019.
B. Garcia-Garcia, T. Bouwmans, A. Rosales-Silva, "Background Subtraction in Real Applications: Challenges, Current Models and Future Directions", Computer Science Review, Volume 35, February 2020.