PHDEC1728 - Adaptive Incremental Deep Learning for Automatic Object Detection and Recognition

   School of Engineering & Computing

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

Click here to search for PhD studentship opportunities
  Dr P Casaseca, Dr Q Wang  No more applications being accepted  Competition Funded PhD Project (European/UK Students Only)

About the Project

Project Ref: PHDEC1728. Deep learning has recently emerged as a game changer in computer vision applications. Convolutional neural networks (CNN) emulate the structure of the visual cortex to automatically detect and recognise small objects (ATD/ATR) with accuracies close to 100%. CNNs automatically extract significant features from the data to enable ATD/ATR with such performance. This is carried out by training a complex network with massive amounts of data. Optimal definition of the network is application dependent, and new considered applications or types of objects will require re-training the network over new training sets. The aim of this project is to better understand this process to enhance and optimize the performance of CNNs in an ATD/ATR context. Specifically, we will address dynamic definition of datasets and network layers to expedite training, and incremental definition of these CNNs to modularly perform different tasks (e.g. ATD or ATR).

This project will pursue the following objectives:

1. To propose adaptive deep learning techniques to automatically define effective ATD/ATR systems for specific applications.
2. To design incremental deep learning strategies based on hierarchical definition of application-specific datasets in order to optimize the training process in a variety of ATD/ATR contexts.
3. To apply the methods and strategies defined above to enable a multi-stage modular ATD/ATR system performing classification and decision-making at different levels.

This project has guaranteed funding from industry (Thales UK, ), the industry-led Innovation Centre for Sensor and Imaging Systems (CENSIS), and UWS. With the objectives above, the project will pave the way for a high quality PhD thesis supported by impacting publications in top journals and conferences. The supervisory team comprises UWS academics and Thales staff. Their expertise in these research topics together with the motivation and involvement of the candidate will guarantee the success of the project.

The research will be carried out within the Artificial Intelligence, Visual Communications and Networks (AVCN) Research Centre ( at the School of Engineering and Computing, UWS. There will also be the opportunity for the student to carry out part of the research within Thales premises. Both centres are equipped with state-of-the-art facilities to provide an adequate framework to develop the project. In particular, AVCNs infrastructure provides massive computational capabilities. Thales on the other hand will provide access to high-end cameras for data acquisition. This will empower their industrial skills in actual product development and improve employability after graduation.

Funding Notes

UWS is an inspiring, vibrant place to study with a growing research community; an important aspect of which is its outstanding and committed research students.

Successful candidates will receive an annual stipend (currently £14,553) per annum for three years and payment of tuition fees (current value £4200). Successful applicants will be expected to contribute up to 6 hours/week to UWS’ academic related activities.


Studentships are open to Home/EU candidates with a first degree in a relevant discipline. Non-EU students can apply, but will not receive the stipend and will be required to pay fees.

How to apply:

Postgraduate Degree by Research Applications should be completed online at

Applications without all relevant documents will not be considered. Please quote the Project Reference Number.


The appointed candidate should have interests and/or expertise in one or more of the following areas:

• Image processing
• Machine learning
• Computer vision
• GPU programming
• Thermal imaging

Experience with one or more of the following technologies is also required: Matlab, Python, C/C++, CUDA

Please contact Dr. Pablo Casaseca ( for further details.