pose estimation

This activity is strongly related to object detection and reconstruction. Pose estimation aims at describing the pose of an object or the relative pose among objects depicted in images or video, in terms of location and orientation. In robotics the awareness of the position and orientation of objects in a scene is sometimes referred to as 6D, where the D stands for degrees of freedom in the pose (or 6DoF). This is an important information when a robotic arm needs to grab and move objects. In extended reality applications the estimation of pose is useful to embed or manipulate virtual objects in the real scene.  

Persons and their body parts are particular categories of objects that are interesting to observe in terms of pose for many applications, at the level of a single person or of groups, for example for studies of behaviors that people display during social interactions.


Recent techniques, which are applied for 6D (six degrees of freedom) pose estimation of rigid objects, rely on deep networks. Considering the important applications in robotics and in augmented reality, we are studying innovative solutions on this task starting from RGB images.

related publication

J. Corsetti, D. Boscaini and F. Poiesi. Revisiting Fully Convolutional Geometric Features for Object 6D Pose Estimation. International Conference on Computer Vision Workshops, October 2023 [arXiv]

M. Bortolon. Classification and pose estimation of objects in a 3D space, Master Thesis in Computer Science, Università di Trento, Italy, 2020/2021, Supervisors: E. Ricci, F. Poiesi   


Simultaneous tracking and head pose estimation enables detailed reporting on attention patterns of people visiting an architectural space or during natural interactions. Main challenges are the very low resolution and non-frontal views of the faces as captured by distant cameras. We developed methods for estimating the head orientation of people from multiple distant cameras integrated with real-time tracking.

selected publications

Y. Yan, E. Ricci, R. Subramanian, G. Liu, O. Lanz and N. Sebe. A Multi-task Learning Framework for Head Pose Estimation under Target Motion, IEEE Transactions on Pattern Analysis and Machine Intelligence ,  38(6):1070-1083, 2016

E. Ricci, J. Varadarajan, R. Subramanian, S. Rota Bulò, N. Ahuja and O. Lanz. Uncovering Interactions and Interactors: Joint Estimation of Head, Body Orientation and F-formations from Surveillance Videos, International Conference on Computer Vision - ICCV,  2015

A.K. Rajagopal, R. Subramanian, E. Ricci, R.L. Vieriu, O. Lanz, R. Kalpathi and N. Sebe. Exploring Transfer Learning Approaches for Head Pose Classification from Multi-view Surveillance Images. International Journal of Computer Vision, 109(1-2):146-167, 2014

Y. Yan, E. Ricci, R. Subramanian, O. Lanz and N. Sebe. No Matter Where You Are: Flexible Graph-guided Multi-task Learning for Multi-view Head Pose Classification Under Target Motion. International Conference on Computer Vision - ICCV, pp. 1177-1184,  2013

A.K. Rajagopal, R. Subramanian, R.L. Vieriu, E. Ricci, O. Lanz, N. Sebe and K. Ramakrishnan. An Adaptation Framework for Head Pose Estimation in Dynamic Multi-view Scenarios. Asian Conference on Computer Vision - ACCV, 2012

M. Voit, N. Gourier, C. Canton-Ferrer, O. Lanz, R. Stiefelhagen and R. Brunelli. Estimation of Head Pose. Computers in the Human Interaction Loop, pp. 33-42, 2009 

O. Lanz and R. Brunelli. Joint Bayesian Tracking of Head Location and Pose from Low-resolution Video. Classification of Events, Activities and Relationships, Evaluation and Workshop - CLEAR, 2007

O. Lanz and R. Brunelli. Dynamic Head Location and Pose from Video. International Conference on Multisensor Fusion and Integration for Intelligent Systems - MFI, pp. 47-52, 2006