Our paper on optical flow estimation is accepted for #ECCV2022 ! We show how to leverage the brightness constancy simplified assumption in a deep augmented model. Joint work with Clément Rambour and Nicolas Thome at CNAM. #computer_vision#optical_flow
Un grand merci pour le prix de thèse Signal Image Vision 2022 (#gdr_isis/#GRETSI /#EEA), et l'accessit au prix de thèse #AFIA! Une belle reconnaissance pour ce projet en deep learning guidé par la physique, appliqué à la prévision d'énergie solaire.
#EDF/#CNAM (@nicolas_thome)
Paper accepted for publication at IEEE PAMI ! We propose to leverage differentiable shape and temporal criteria for training deep time series forecasting models. #IEEECS
Extending two #NeurIPS papers: DILATE (#neurips2019), STRIPE (#neurips2020)
Paper: https://t.co/QjCSqUPGLZ
Our paper APHYNITY was accepted for an Oral at #ICLR2021 ! How to ensure a meaningful decomposition between physical and machine learning models? We invite you to check the paper: https://t.co/dPALEH02WW
Joint work with collegues from Sorbonne University and CNAM Paris
Our CVPR 2020 paper PhyDNet on unsupervised video prediction is now available on Arxiv (https://t.co/lK5SK34OMX) and code coming soon on Github(https://t.co/S1kqvY1RBP) #CVPR2020#CVPR
Our paper on video prediction with the introduction of physical priors from partial differential equations was accepted for #CVPR2020 ! Will soon be available on Arxiv !
Our paper "Shape and Time Distortion Loss for Training Deep Time Series Forecasting Models" accepted for NeurIPS 2019 conf now on Arxiv ! See you in Vancouver
#NeurIPS#NeurIPS2019
https://t.co/dTBgl1XyCE