Project highlights

  • Development of rapid and automated scene anomaly detection with a focus on identifying differences in biota and anthropogenic features.  
  • Utilising the above, specific near-realtime automated identification and measurements of benthic anthropogenic debris, scallop detection and nephrops norvegicus burrow identification and network detection. 
  • AI-powered marine monitoring 

Overview

This PhD project focuses on advancing computer vision and edge-AI technology for real-time marine monitoring. In collaboration with CEFAS (the Centre for Environment, Fisheries, and Aquaculture Science), a global leader in marine science, the project will develop scalable, low-cost embedded vision systems to analyze marine biodiversity and detect anthropogenic debris. The core challenge is creating robust, real-time edge-AI algorithms capable of accurately classifying diverse marine species and debris under complex, dynamic underwater conditions, while minimizing reliance on expensive cloud infrastructure. 

The project will build on a working prototype, the Neural Network Enhanced Marine Observation system (NNEMO), a low-cost, shallow-water, edge-AI-enabled spatial camera system designed to meet multiple needs in marine biodiversity monitoring. The current system successfully classifies and counts key indicator species, such as sea pens, and the project aims to extend its capabilities to include benthic anthropogenic debris classification using in-house trained deep learning models deployed onboard. 

The demand for such a low-cost system stems from the need to increase efficiency in marine monitoring, where the high cost of offshore and nearshore vessel deployment, along with the laborious analysis of imagery, slows data collection and monitoring. Furthermore, existing computer vision solutions often depend on cloud computing infrastructure and require specialized expertise. A scalable, embedded computer vision system that can analyze imagery in real time offers substantial value by enabling more accurate assessments of nearshore vegetation, shellfish stocks, anthropogenic debris, and epibenthic biodiversity. 

This project aims to address these challenges and deliver an innovative, practical solution for real-time marine monitoring. The objectives of the project include: 

Objective 1) Calibrate spatial AI algorithm, correcting for seawater and lens distortion.  

Objective 2) Develop a pipeline to estimate (using stereo derived image depth calculations) organism and anthropogenic debris volumes from image depth data. 

Objective 3) Develop scallop detection and measuring algorithms, alongside anthropogenic benthic debris and nephrops norvegicus burrow identification and network detection 

Objective 4) Integrate the developed algorithms with NNEMO system and conduct testing.  

CENTA Flagship

This is a CENTA Flagship Project

Case funding

This project is suitable for CASE funding

Host

Loughborough University

Theme

  • Climate and Environmental Sustainability
  • Organisms and Ecosystems

Supervisors

Project investigator

Co-investigators

How to apply

Methodology

The PhD candidate will utilize the Depth AI Python library and SDK, alongside OpenCV and deep learning platforms such as PyTorch, in a tank-testing environment to establish stereo calibration for the NNEMO platform’s native image depth model. These calibrated spatial models will be deployed during nephrops norvegicus and Scallop surveys in 2025 to capture depth-enabled seabed footage. Additionally, the spatial imagery collected from these surveys, combined with data from marine protected area surveys, will support the development of models for benthic anthropogenic debris detection, scallop identification and tracking, and nephrops norvegicus burrow identification and network mapping. 

A key objective will be to develop a pipeline for estimating organism volumes using stereo-derived image depth calculations. The models and algorithms created will then be integrated with other NNEMO platform functionalities to facilitate real-time testing. 

Training and skills

DRs will be awarded CENTA Training Credits (CTCs) for participation in CENTA-provided and ‘free choice’ external training. One CTC can be earned per 3 hours training, and DRs must accrue 100 CTCs across the three and a half years of their PhD.  

In addition to CENTA training opportunities, this PhD project will provide extensive training in artificial intelligence, computer vision, and embedded AI, with a specific focus on their application in real-time marine monitoring systems. The candidate will gain advanced skills in developing deep learning and computer vision models for underwater object detection and classification. Additionally, they will acquire expertise in robotics and embedded systems, focusing on edge computing for data collection and analysis. Through interdisciplinary collaboration with CEFAS, the candidate will gain related knowledge in marine science and develop practical skills in deploying AI technologies in challenging environments such as marine applications. 

Partners and collaboration

Cefas, an executive agency of Defra with over 500 staff and 80 PhD students, is a global leader in marine science. Cefas helps to shape and implement policy through its internationally-renowned science and collaborative relationships. 

The Department of Computer Science at Loughborough University has a world reputation for its impactful research, achieving 100% world-leading or internationally excellent research impact in REF2021. Lead supervisor Professor Qinggang Meng, an expert in AI, robotics, and computer vision, along with complementary expertise from Cefas, will provide strong support for this PhD project. The team also has excellent facilities including 8-A100 and 4-A100 GPU servers.  

Further details

For further information about this project, please contact Professor Qinggang Meng ([email protected]).

To apply to this project:  

  • You must include a CV with the names of at least two referees (preferably three) who can comment on your academic abilities.  
  • Please submit your application and complete the host institution application process via: https://www.lboro.ac.uk/study/postgraduate/apply/research-applications/   The CENTA Studentship Application Form 2025 and CV, along with other supporting documents required by Loughborough University, can be uploaded at Section 10 “Supporting Documents” of the online portal.  Under Section 4 “Programme Selection” the proposed study centre is Central England NERC Training Alliance.  Please quote CENTA 2025-LU8 when completing the application form. 
  • For further enquiries about the application process, please contact the School of Social Sciences & Humanities ([email protected]). 

Applications must be submitted by 23:59 GMT on Wednesday 8th January 2025.  

Possible timeline

Year 1

Literature Review,  and Initial feasibility studies 

  • Months 1-3: Conduct an extensive literature review on AI, and computer vision and their applications on marine monitoring, and learn knowledge on images of anthropogenic benthic debris and nephrops norvegicus.   Define research questions and goals. 
  • Months 4-6: Identify and gather relevant datasets. Get familiar with the NNEMO  system. Set up the hardware and software environment.  
  • Months 7-12: Begin initial data analysis and develop preliminary models. Conduct feasibility studies.  

In year1, the student also needs to Complete any required training (courses, workshops). 

Year 2

Algorithms and models for scallop detection, nephrops  norvegicus burrow identification, and benthic anthropogenic debris measurement  

  • Months 13-15: Refine models or algorithms for scallop detection, iterating based on initial findings. Collect new data if necessary. 
  • Months 16-19: Develop nephrops  norvegicus burrow identification and network detection algorithms 
  • Months 20-22: Develop deep learning algorithms for benthic anthropogenic debris measurement  
  • Months 23-24:  Prepare paper (s)  for conferences. 

Year 3

Implement algorithms on an embedded AI board and conduct final experiments, result analysis, and thesis writing 

  • Months 25-26: Integrate the algorithms with the NNEMO  system  
  • Months 27-29: Complete final experiments and refine models and algorithms according to the experimental data. 
  • Months 30-31: Conduct data analysis and comparison with the state of the art methods, write a conference paper.  
  • Months 32-36: Write a journal paper, and finish the PhD thesis writing up.  

Further reading

Greene, H.G., Brown, C.J., Harris, P.T. and Picard, K. (2023). Mapping anthropogenic impacts on marine benthic habits, Continental Shelf Research, Vol 269, 105142 

Naseer, A., Baro, E.N., Khan, S.D. and Vila, Y., (2022). A Novel Detection Refinement Technique for Accurate Identification of Nephrops norvegicus Burrows in Underwater Imagery, Sensors 22, no. 12: 4441. https://doi.org/10.3390/s22124441 

Shenming Qu, Cui, C., Duan, J., Lu, Y. and Pang, Z. (2024). Underwater small target detection under YOLOv8-LA model, Scientific  Report 14, 16108 

Song, T., Pang, C., Hou, B., Xu, G., Xue, J., Sun H. and Meng, F. (2023). A review of artificial intelligence in marine science, Frontiers in Earth Science, Volume 11,  https://doi.org/10.3389/feart.2023.1090185. 

Zhou, Y., Wang, L.B., Rocco, R. and Meng, Q (2022) Discovering Unknowns: Context-enhanced Anomaly Detection for Curiosity-driven Autonomous Underwater Exploration, Pattern Recognition, Vol.131, 108860. DOI: 10.1016/j.patcog.2022.108860.