Welcome to the PERCEPTION’s website at INRIA Grenoble Rhone-Alpes

PERCEPTION is a research group associated with INRIA, Centre de Recherche INRIA Grenoble Rhone-Alpes, and with Laboratoire Jean Kuntzmann (LJK). The scientific objective of the group is to map images and videos onto 3-D visual representations. More specifically we address the following broad topics:

  Building rich 3-D representations using geometry, texture, and photometry in combination with multiple cameras;
  Modeling articulated objects (recognition, tracking, matching, motion capture, motion segmentation, action and gesture interpretation);
  Biologically perception from the viewpoint of computer vision, and
  Integration between vision and other sensing modalities, in particular audio-visual fusion.

We study the mathematical models that are needed to extract 3-D content from the light stimuli (images) and we devise methods, algorithms, and software packages that best implement these models. We also investigate the biological/cognitive processes that govern the interactions between high-level representations and the sensory inputs, we attempt to describe these processes as mathematical and computational models, and we implement them as working computer programs.

More...

Recent Publications

• Conjugate Mixture Models for Clustering Multimodal Data
  Vasil Khalidov, Florence Forbes, Radu P. Horaud
• Disparity and Normal Estimation through Alternating Maximization
  Ramya Narasimha, Elise Arnaud, Florence Forbes, Radu P. Horaud
• Learning Shape Segmentation Using Constrained Spectral Clustering and Probabilistic Label Transfer
  Avinash Sharma, Etienne von Lavante, Radu P. Horaud
• Shape Matching Based on Diffusion Embedding and on Mutual Isometric Consistency
  Avinash Sharma, Radu P. Horaud
• Satellite Image Registration for Attitude Estimation with a Constrained Polynomial Model
  Régis Perrier, Elise Arnaud, Peter Sturm, Mathias Ortner
• Image-Based Satellite Attitude Estimation
  Régis Perrier, Elise Arnaud, Peter Sturm, Mathias Ortner
• Bas-Relief Ambiguity Reduction in Shape from Shadowgrams
  Jamil Draréni, Sébastien Roy, Peter Sturm
• A global camera network calibration method with Linear Programming
  Jérôme Courchay, Arnak Dalalyan, Renaud Keriven, Peter Sturm
• Plane-Based Calibration for Linear Cameras
  Jamil Draréni, Sébastien Roy, Peter Sturm
• One-shot scanning method using an uncalibrated projector and camera system
  Hiroshi Kawasaki, Ryusuke Sagawa, Yasushi Yagi, Ryo Furukawa, Naoki Asada, Peter Sturm
• More...