Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Computer vision methods for irregularity detection in fish nets
Fish Farming industry is facing several challenges, with one of the major challenges being related to objective inspection of fish cages to detect irregularities such as holes, biofouling condition [1]. Earlier studies showed that 41% of the escapees from fish farms in Norway are caused by holes in fish cages [2] and the biofouling prevention is crucial to preserve good conditions for the fish growth [1].
Keywords: Net Inspection, Computer Vision, Irregularity Detection
Inspection of fish cage is essential to detect and prevent the possible escape of farmed fish with the overall goal to minimize the risk of any negative impact to wild fish. In addition, frequent inspection and identification of biofouling conditions are highly relevant for the operations planning for net cleaning operations. Many industry actors offering net inspection and cleaning services using underwater vehicles have shown interest in such solutions. Today the captured videos are manually inspected for holes and irregularities. Earlier studies in this domain targeted the development of different methods utilizing computer vision methods, lack robustness and/or real-time performance [3][4]. This thesis aims to investigate and develop methods to detect and locate the irregularities in fish nets utilizing visual datasets obtained from stereo camera attached on BlueROV2. The proposed approach will aim near real-time performance suited for autonomous operations with underwater vehicles in industrial scale fish farms. In addition to access to relevant already available datasets from stereo camera, there is potential for dedicated datasets recording from SINTEF ACE full scale aquaculture lab [5].
References:
[1] Martin Føre et al. “Precision fish farming: A new framework to improve production in aquaculture”. Biosystems Engineering, 2017
[2] Why do the fish escape? https://www.youtube.com/watch?v=cKSLnneC8RA
[3] Yip, Mauhing, et al. "Robust Hole-Detection in Triangular Meshes Irrespective of the Presence of Singular Vertices, 2023
[4] C Schellewald and A Stahl, Irregularity detection in net pens exploiting Computer Vision - IFAC-PapersOnLine, 2022
[5] SINTEF ACE. https://www.sintef.no/en/all-laboratories/ace/
Inspection of fish cage is essential to detect and prevent the possible escape of farmed fish with the overall goal to minimize the risk of any negative impact to wild fish. In addition, frequent inspection and identification of biofouling conditions are highly relevant for the operations planning for net cleaning operations. Many industry actors offering net inspection and cleaning services using underwater vehicles have shown interest in such solutions. Today the captured videos are manually inspected for holes and irregularities. Earlier studies in this domain targeted the development of different methods utilizing computer vision methods, lack robustness and/or real-time performance [3][4]. This thesis aims to investigate and develop methods to detect and locate the irregularities in fish nets utilizing visual datasets obtained from stereo camera attached on BlueROV2. The proposed approach will aim near real-time performance suited for autonomous operations with underwater vehicles in industrial scale fish farms. In addition to access to relevant already available datasets from stereo camera, there is potential for dedicated datasets recording from SINTEF ACE full scale aquaculture lab [5].
References: [1] Martin Føre et al. “Precision fish farming: A new framework to improve production in aquaculture”. Biosystems Engineering, 2017 [2] Why do the fish escape? https://www.youtube.com/watch?v=cKSLnneC8RA [3] Yip, Mauhing, et al. "Robust Hole-Detection in Triangular Meshes Irrespective of the Presence of Singular Vertices, 2023 [4] C Schellewald and A Stahl, Irregularity detection in net pens exploiting Computer Vision - IFAC-PapersOnLine, 2022 [5] SINTEF ACE. https://www.sintef.no/en/all-laboratories/ace/
- Perform a literature review on current status and challenges in fish farms and computer vision methods for irregularity detections (e.g., hole detection, biofouling conditions, etc)
- Implement, test, compare and assess the performance of different suited methods for irregularity detection with the aim to increase robustness and reach near real-time performance
- Identify and report on irregularities utilizing the provided datasets from industrial scale fish farms
- Perform a literature review on current status and challenges in fish farms and computer vision methods for irregularity detections (e.g., hole detection, biofouling conditions, etc) - Implement, test, compare and assess the performance of different suited methods for irregularity detection with the aim to increase robustness and reach near real-time performance - Identify and report on irregularities utilizing the provided datasets from industrial scale fish farms
- Highly motivated student
- Experience with computer vision
- Experience with C/C++ and/or Python programming is desired
- Highly motivated student - Experience with computer vision - Experience with C/C++ and/or Python programming is desired