Current Work

My research is at the intersection between modern Computer Vision and Variational Inference. My goal is to combine deep-learning techniques with scalable inference algorithm in order to develop new structured learning methods.

I am currently focusing on learning Mean-Field Inference algorithm for Multi-Camera Pedestrian Detection and occlusion reasoning.

Recent Publications

MF
Deep-Occlusion Reasoning for Multi-Camera Multi-People detection

Pierre Baqué, François Fleuret, Pascal Fua – ICCV 2017


MF

Multi-Modal Mean-Fields via Cardinality-Based Clamping

Pierre Baqué, François Fleuret, Pascal Fua – CVPR 2017



MF
Principled Parallel Mean-Field Inference for Discrete Random Fields

Pierre Baqué, Timur Bagautdinov, François Fleuret, Pascal Fua – CVPR 2016


Prox

Kullback-Leibler Proximal Variational Inference​

Mohammad E. Khan, Pierre Baqué, François Fleuret, Pascal Fua – NIPS 2015