Continuous Deep Human Pose Estimation For Video Sequences

  • Continuous Deep Human Pose Estimation For Video Sequences

School of Electronics, Electrical Engineering and Computer Science
& ECIT Global Research Institute

Proposed Project Title: Continuous Deep Human Pose Estimation For Video Sequences

Principal Supervisor:   Jesus Martinez del Rincon                   Second Supervisor: Dr. Neil Robertson

Project Description:

Human pose recovery is one of the most challenging research areas in computer vision. Successful estimations of the human pose provide information and simplify further tasks such as activity recognition or behaviour analysis, and therefore, it can benefit a wide range of industrial sectors such as video surveillance, physical security or sport performance enhancement.

While constant steps forward have been made in this field, the 3D pose estimation from monocular images remains elusive. Recent advances in deep learning and their application to pose estimation have allowed to better reconstruct the 3D skeleton from static images. While impressive results have been achieved in certain images and poses, results tend to be inconsistent and inaccurate in upper and lower limbs. This is due to the inherent ambiguity of the 2D-to-3D projection and the presence of self-occlusions. As a result, when these techniques are applied to full video sequences the reconstructed skeletons do not keep continuity and suffer from jittering, making impossible to understand the performed motion or activity.

The aim of this project is to investigate different methods that allow introducing and exploiting the temporal information  within deep learning architecture for an enhance pose estimation in continuous vide sequences.


Objectives:

  • To develop new deep learning architectures for pose estimation in video sequences that allow feeding in and feeding back temporal information such as  the previous or the most likely poses.
  • To investigate the suitability of recurrent  layers such as LSTM to learn and preserve temporal pose sequentiality.
  • To evaluate the performance of the developed methodologies against occlusions and the role of artificial data augmentation for improving robustness.

Academic Requirements:

Students entering the programme will normally be required to have a 2.1 BSc/BEng in Computer Science, Electrical and Electronic Engineering, or a maths based engineering or physical science degree, or equivalent qualification recognised by the University. Students holding an appropriate MEng or MSc (Software conversion) will normally be required to have a 2.1 or commendation (distinction) respectively. Furthermore, additional criteria may be applied. All applicants must have significant mathematical and programming experience.

GENERAL INFORMATION:

This 4 year PhD studentship, potentially funded by the Department for Employment and Learning (DEL), commences on 1 October 2018.

Eligibility for both fees and maintenance depends on the applicants being either an ordinary UK resident or those EU residents who have lived permanently in the UK for the 3 years immediately preceding the start of the studentship. Non UK residents who hold EU residency may also apply but if successful may receive fees only.

Applicants should apply electronically through the Queen’s online application portal at: https://dap.qub.ac.uk/portal/

Further information available at:  http://www.qub.ac.uk/schools/eeecs/StudyattheSchool/PhDProgrammes



Contact details

Supervisor Name: Jesus Martinez del Rincon                              Tel: +44 (0)28 9097 1779
QUB Address:  ECIT Building, Queen’s Road                               Email: j.martinez-del-rincon@qub.ac.uk