FindAPhD Weekly PhD Newsletter | JOIN NOW FindAPhD Weekly PhD Newsletter | JOIN NOW

Identification of the type of the debris in marine ecosystem using explainable deep learning methods (Advert Reference: RDF22/EE/CIS/KURT)

   Faculty of Engineering and Environment

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

Click here to search for PhD studentship opportunities
  Dr Zeyneb Kurt, Prof Wai Lok Woo  No more applications being accepted  Competition Funded PhD Project (Students Worldwide)

About the Project

Debris in seas and oceans can impact the marine ecosystem destructively, since many marine animals mistake foreign materials for food and try to ingest them, causing obstruction of their gastrointestinal tract and resulting in fatality. Less lethal, yet detrimental effects of marine pollution may lead to poisoning of the animals by plastic fragments contaminated with polychlorinated biphenyl (PCB) or heavy metals, and cause epigenetic perturbations in their reproductive systems. Marine debris heavily impacts not only the environment, but also human health. Nano plastics have been discovered to make their way into human anatomy via marine-originated salt, sugar, fish, and even water itself; and it is estimated that we consume a diverse variety of toxic plastic fragments which equate to approximately a credit card per week. Consequently, this dramatically boosts the global disease burden (in terms of healthy life years lost and premature mortality).

Impact: Hence, not only the oceans or planet but also our lives, which are highly interconnected, would be benefiting from an automated system that can identify marine debris.

Gaps: Current applications have not considered protection of wildlife as they frequently misclassify some marine plants or animals as foreign materials (for instance, jellyfish can be mistaken for a plastic bag), and in a fully automated system this can cause mistakenly removal of animals or plants as if they were debris, which could deteriorate the sensitive equilibrium of the underwater ecosystems. Most models can only detect floating debris or the ones close to the surface, but debris located at different depths, or different density or lightening levels of the sea can be missed frequently. Additionally, at most three or four categories of debris were considered thus far (such as plastic bottles, straws, buckets), which corresponds to a tiny portion of marine debris (e.g. straws only account for 0.03%). Hence, an image database including diverse types of debris, as well as a natural marine environment with plants and animals (without debris), within a variety of conditions (e.g. different sea depth, density, light levels) is needed to train such models. No such publicly available image database exists with proper labelling. Our aim is to distinguish the debris-free wildlife environment images from the ones including debris, and also classifying the type of debris accurately by curating a large-scale image repository and fine-tuning the commonly used convolutional neural network (CNN) architectures.


  • (i) To create a large image repository of marine life by contacting various organisations and conservationists;
  • (ii) To label the images with the type of debris by extending the number of categories - different from the previous studies which limit it to four;
  • (iii) To apply data augmentation and provide an approximate uniform distribution between the classes.
  • (iv) To fine-tune the parameters of various CNN models (e.g. VGG, Inception, ResNet) using 80% of the images, optimizing the architecture, while avoiding the overfitting of models.
  • (v) To test the re-trained models with 20% of the images in the database.
  • (vi) To add explainable functionality to the CNN models so that the decision is transparent to the users.

The Principal Supervisor for this project is Dr. Zeyneb Kurt.

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.
  • Applicants cannot apply for this funding if currently engaged in Doctoral study at Northumbria or elsewhere or if they have previously been awarded a PhD.

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. RDF22/…) will not be considered.

Deadline for applications: 18 February 2022

Start Date: 1 October 2022

Northumbria University takes pride in, and values, the quality and diversity of our staff and students. We welcome applications from all members of the community.

Funding Notes

Each studentship supports a full stipend, paid for three years at RCUK rates (for 2021/22 full-time study this is £15,609 per year) and full tuition fees. UK and international (including EU) candidates may apply.
Studentships are available for applicants who wish to study on a part-time basis over 5 years (0.6 FTE, stipend £9,365 per year and full tuition fees) in combination with work or personal responsibilities.
Please also read the full funding notes which include advice for international and part-time applicants.


Kurt Z, Barrere-Cain R, Laguardia J, et al. "Tissue-specific pathways and networks underlying sexual dimorphism in non-alcoholic fatty liver disease", Biology of Sex Differences 2018, vol. 9:46. doi: 10.1186/s13293-018-0205-7.
Krishnan KC*, Kurt Z*, Barrere-Cain R, et al. “Integration of Multi-omics Data from Mouse Diversity Panel Highlights Mitochondrial Dysfunction in Non-Alcoholic Fatty Liver Disease”, Cell Systems 2017. doi: 10.1016/j.cels.2017.12.006. *Shared (co-first) authors.
Erdogan C, Kurt Z, Diri B, "Estimation of the proteomic cancer co-expression subnetworks by using association estimators", PLOS ONE 2017, doi:10.1371/journal.pone.0188016.
Kurt Z, Yavuz S, "Improvement of the measurement update step of EKF-SLAM", IEEE 16th International Conference on Intelligent Engineering Systems (INES), 2012, pp. 61-65, doi: 10.1109/INES.2012.6249803.
Sakarkaya M, Yanbol F, Kurt Z "Comparison of several classification algorithms for gender recognition from face images" IEEE 16th International Conference on Intelligent Engineering Systems (INES), 2012, pp. 97-101, doi: 10.1109/INES.2012.6249810.
Kurt Z, Turkmen HI, Karsligil ME. “Linear Discriminant Analysis in Ottoman Alphabet Character Recognition”. In: Mastorakis N., Mladenov V., Kontargyri V. (eds) Proceedings of the European Computing Conference. Lecture Notes in Electrical Engineering, vol 28. Springer, Boston, MA.
B. Hu, B. Gao, W.L. Woo, “A Lightweight Spatial and Temporal Multi-feature Fusion Linked Self-Attention Network for Defect Detection,” IEEE Trans. Image Processing, vol. 30, pp. 472-486, 2021
J. Ahmed, B. Gao, W.L. Woo, “Sparse Low-Rank Tensor Decomposition for Metal Defect Detection Using Thermographic Imaging Diagnostics,” IEEE Trans. on Industrial Informatics, vol. 17, no. 3, pp. 1810-1820, 2021
Juniad Ahmed, Bin Gao, W.L. Woo, “Wavelet Integrated Alternating Sparse Dictionary Matrix Decomposition in Thermal Imaging CFRP Defect Detection,” IEEE Trans. on Industrial Informatics, vol. 5, no. 7, pp. 4033-4043, 2019
PhD saved successfully
View saved PhDs