Semantic segmentation is the problem of identifying pixels that constitute different objects in an image. One of the most prominent applications of semantic segmentation is navigation of autonomous machines such as self-driving cars, drones, etc. Self-driving vehicles vehicles are required to detect other vehicles and pedestrians present in images captured using on-board cameras. This is imperative however problem is complicated by the view-angle dependent differences in the appearance of a given object. Furthermore, in the life-critical application of self-driving cars, as a necessity this operation must be executed repeatedly in a millisecond time-scale with high accuracy.
In the last decade, deep neural network based techniques have resulted in significant improvements in this task. These networks expose an image to multiple layers of processing in order to detect different objects in the image and assign each pixel to a particular object. These techniques utilize powerful Graphical Processing Units (GPU) in order to achieve fast response times. However, the high energy requirement and additional weight associated with GPUs make them less suitable for on-board processing in self-driving machines. There has been a significant bottleneck with regards to deployment of these deep learning techniques as an on-board solution.
This project aims to develop deep learning techniques that can achieve current state-of-the-art performance but have orders of magnitude lower energy footprint while maintaining the low latency exhibited by deep networks. In particular, the project will utilize deep spiking neural networks which are inspired by the way information is processed in the brain. These networks employ neurons that process and transmit information using binary events in time called spikes. The event-based processing employed by these networks render them highly energy-efficient in comparison to traditional deep neural networks.
Additionally, the inherent ability of these networks to represent temporal information allows them to naturally encode temporally correlated information received by vision sensors in real world applications. As a result, these network support a straightforward method to use segmentation techniques for real-time tracking of objects in data obtained from vision sensors.
In this project, we will exploit these properties of spiking neural networks for development of techniques for real time segmentation of pedestrian, vehicles and other relevant objects in the domain of self-driving cars.
Applicants should hold, or expect to obtain, a First or Upper Second Class Honours Degree in a subject relevant to the proposed area of study.
We may also consider applications from those who hold equivalent qualifications, for example, a Lower Second Class Honours Degree plus a Master’s Degree with Distinction.
In exceptional circumstances, the University may consider a portfolio of evidence from applicants who have appropriate professional experience which is equivalent to the learning outcomes of an Honours degree in lieu of academic qualifications.
If the University receives a large number of applicants for the project, the following desirable criteria may be applied to shortlist applicants for interview.
The University offers the following levels of support:
The following scholarship options are available to applicants worldwide:
These scholarships will cover full-time PhD tuition fees for three years (subject to satisfactory academic performance) and will provide a £900 per annum research training support grant (RTSG) to help support the PhD researcher.
Applicants who already hold a doctoral degree or who have been registered on a programme of research leading to the award of a doctoral degree on a full-time basis for more than one year (or part-time equivalent) are NOT eligible to apply for an award.
Please note: you will automatically be entered into the competition for the Full Award, unless you state otherwise in your application.
The scholarship will cover tuition fees at the Home rate and a maintenance allowance of £19,000 (tbc) per annum for three years (subject to satisfactory academic performance).
This scholarship also comes with £900 per annum for three years as a research training support grant (RTSG) allocation to help support the PhD researcher.
Due consideration should be given to financing your studies. Further information on cost of living
1. Liang, Ming, et al. "PnPNet: End-to-End Perception and Prediction with Tracking in the Loop." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020.
2. Shrestha, Sumit B., and Garrick Orchard. "Slayer: Spike layer error reassignment in time." Advances in Neural Information Processing Systems 31 (2018): 1412-1421.
3. Dora, Shirin, Cyriel Pennartz, and Sander Bohte. "A deep predictive coding network for learning latent representations." (2018).
Submission deadline
Friday 5 February 2021
12:00AM
Interview Date
25 March 2021
Preferred student start date
Mid September 2021