Durham University Featured PhD Programmes
Health Data Research UK Featured PhD Programmes
Centre for Genomic Regulation (CRG) Featured PhD Programmes

Developing Robust Visual Recognition Frameworks using Synthetic Data (Application Ref: SF19/EE/CIS/HO)

Faculty of Engineering and Environment

This project is no longer listed on FindAPhD.com and may not be available.

Click here to search FindAPhD.com for PhD studentship opportunities
Dr Edmond Ho , Dr H Shum Applications accepted all year round Self-Funded PhD Students Only

About the Project

Deep Learning is one of the major research themes across different areas in scientific research. With the promising performance and results presented over the last few years, deep learning algorithms are dominating a lot of data-driven tasks in visual recognition. However, one of the pre-requisites of deploying deep learning algorithms is to have a significant volume of data for the training process. Collecting data can be costly, in terms of both labour cost and time. In this research, we are going to investigate how synthetic data can be used in training deep learning frameworks for handling visual data captured from the real world.

With the advancement of Generative adversarial network (GAN), realistic visual data can be generated with a relatively smaller amount of training data. In addition, the dual-GAN structures such as Dual-GAN, Cycle-GAN and augmented Cycle-GAN demonstrated superior performance in transferring styles between different visual data. For example, transferring the style from sketches to photos. Inspired by this powerful feature, this project will investigate the feasibility of transferring the style of synthesized data (e.g. generated from 3D computer graphics techniques) to realistic photos. By this, unlimited training data can be generated for training deep learning frameworks. The initial experiments will be focusing on popular visual recognition tasks such as pose estimation, face recognition and pose reconstruction.

The supervisory team has extensive experience in Computer Graphics, Computer Vision and Machine Learning (http://www.edho.net). The PhD candidate will receive training in the aforementioned areas, as well as deep learning programming. The team has on-going collaborations with professional users such as NHS hospitals and City councils in North East England, which can further extend the impact of this research to solve real-world problems on the society in the future.

This project is supervised by Dr. Edmond Ho.

Eligibility and How to Apply:

Please note eligibility requirement:
• Academic excellence of the proposed student i.e. 2:1 (or equivalent GPA from non-UK universities [preference for 1st class honours]); or a Masters (preference for Merit or above); or APEL evidence of substantial practitioner achievement.
• Appropriate IELTS score, if required.

For further details of how to apply, entry requirements and the application form, see

Please note: Applications that do not include a research proposal of approximately 1,000 words (not a copy of the advert), or that do not include the advert reference (e.g. SF19/EE/CIS/HO) will not be considered.

Start Date: 1 March 2020 or 1 October 2020

Northumbria University takes pride in, and values, the quality and diversity of our staff. We welcome applications from all members of the community. The University holds an Athena SWAN Bronze award in recognition of our commitment to improving employment practices for the advancement of gender equality and is a member of the Euraxess network, which delivers information and support to professional researchers.

Funding Notes

This is an unfunded research project.


Refereed Journal Papers

1 H Wang, E S L Ho, H P H Shum, Z Zhu, “Spatio-temporal Manifold Learning for Human Motions via Long-horizon Modeling”, IEEE transactions on visualization and computer graphics, accepted in 2019.

2 Y Shen, L Yang, E S L Ho, H P H Shum, “Interaction-based Human Activity Comparison”, IEEE transactions on visualization and computer graphics, accepted in 2019.

3 D Sakkos, E S L Ho, H P H Shum, “Illumination-aware Multi-task GANs for Foreground Segmentation”, IEEE Access, vol 7, page 10976-10986, Jan 2019.

4 K Yin, H Huang, E S L Ho, H Wang, T Komura, D Cohen-Or, H Zhang, “A Sampling Approach to Generating Closely Interacting 3D Pose-pairs from 2D Annotations”, IEEE Transactions on Visualization and Computer Graphics, accepted in 2018.

5 W Rueangsirarak, J Zhang, N Aslam, E S L Ho, H P H Shum, “Automatic Musculoskeletal and Neurological Disorder Diagnosis with Relative Joint Displacement from Human Gait”, IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol 26(12), page 2387-2396, Nov 2018.

6 E S L Ho, J C P Chan, H P H Shum, Y M Cheung, P C Yuen, “Improving Posture Classification Accuracy for Depth Sensor-based Human Activity Monitoring in Smart Environments”, Computer Vision and Image Understanding, vol 148, page 97-110, July 2016.

Refereed Conference Papers

1 D Sakkos, E S L Ho, H P H Shum, “Illumination-Based Data Augmentation for Robust Background Subtraction”, Proceedings SKIMA 2019, Aug 2019. (* Best Paper Award)

2 K D McCay, E S L Ho, C Marcroft, N D Embleton, “Establishing Pose Based Features Using Histograms for the Detection of Abnormal Infant Movements”, Proceedings of EMBC, July 2019.

3 S Xu, E S L Ho, N Aslam, H P H Shum, “Unsupervised Abnormal Behaviour Detection with Overhead Crowd Video”, Proceedings of SKIMA 2017.

Full publication list: http://www.edho.net/publication.php

FindAPhD. Copyright 2005-2020
All rights reserved.