For EU member states to meet the requirements of the Marine Strategy Framework Directive and the reformed Common Fisheries Policy, it will be necessary to improve data collection related to many fisheries that are at present subject to relatively little monitoring or scientific research. This study evaluated the use of on-board camera systems to collect data from Cancer pagurus and Homarus gammarus fisheries. We evaluated the reliability of the hardware and its ability to collect images of sufficient accuracy and precision compared with using on-board observers. Fishers and on-board observers passed animals removed from traps across a defined area. The relationship between the in situ and predicted measurements of carapace length of lobsters or carapace width (CW) of crabs was investigated. The mean difference between the predicted and real crab measurements was −0.853 mm with a standard error of 0.378 mm. Suggesting that the model tends to underestimate the real CW slightly. The mean difference between predicted and real data for lobsters was 0.085 mm with a standard error of 0.208 mm. Sex allocation for crabs based on video images was 100% accurate. All male lobsters were correctly assigned. For lobsters >86 mm in length, the correct female sex allocation was 100% accurate. For smaller lobsters, the accuracy of sex allocation decreased to a low of 51% in lobsters <70 mm. Camera systems were found to be a suitable method for collecting data on the size and sex of crabs and lobsters. The error attributable to using video data rather than manual measurement was less than 3 mm, which is sufficient to detect growth increments in these species. The requirements to collect basic species data are increasing and the ability to do so without on-board observers will reduce the cost implications of these requirements. Future computer automation of image extraction and measurements will increase the application of video systems for data collection.

Connectivity between populations is important when considering conservation or the management of exploitation of vulnerable species. We investigated how populations of a broadcast‐spawning marine species (scallop, Pecten maximus) that occur in discrete geographic locations were connected to each other. Population genetic insights were related to the outputs from a three‐dimensional hydrodynamic model implemented with scallop larval behaviour to understand the extent to which these areas were linked by oceanographic processes and how this was altered by season and two contrasting years that had strongly different average temperature records (warm vs cold) to provide contrasting oceanographic conditions. Our results span from regional to shelf scale. Connectivity was high at a regional level (e.g. northern Irish Sea), but lower at scales >100 km between sites. Some localities were possibly isolated thus dependent on self‐recruitment to sustain local populations. Seasonal timing of spawning and inter‐annual fluctuations in seawater temperature influenced connectivity patterns, and hence will affect spatial recruitment. Summer rather than spring spawning increased connectivity among some populations, due to the seasonal strengthening of temperature‐driven currents. Furthermore, the warm year resulted in higher levels of modelled connectivity than the cold year. The combination of genetic and oceanographic approaches provided valuable insights into the structure and connectivity at a continental shelf scale. This insight provides a powerful basis for defining conservation management units and the appropriate scale for spatial management. Temporal fluctuations in temperature impact upon variability in connectivity, suggesting that future work should account for ocean warming when investigating population resilience.

This work investigates the application of Computer Vision to the problem of the automated counting and measuring of crabs and lobsters onboard fishing boats. The aim is to provide catch count and measurement data for these key commercial crustacean species. This can provide vital input data for stock assessment models, to enable the sustainable management of these species. The hardware system is required to be low-cost, have low-power usage, be waterproof, available (given current chip shortages), and able to avoid over-heating. The selected hardware is based on a Raspberry Pi 3A+ contained in a custom waterproof housing. This hardware places challenging limitations on the options for processing the incoming video, with many popular deep learning frameworks (even light-weight versions) unable to load or run given the limited computational resources. The problem can be broken into several steps: (1) Identifying the portions of the video that contain each individual animal; (2) Selecting a set of representative frames for each animal, e.g, lobsters must be viewed from the top and underside; (3) Detecting the animal within the frame so that the image can be cropped to the region of interest; (4) Detecting keypoints on each animal; and (5) Inferring measurements from the keypoint data. In this work, we develop a pipeline that addresses these steps, including a key novel solution to frame selection in video streams that uses classification, temporal segmentation, smoothing techniques and frame quality estimation. The developed pipeline is able to operate on the target low-power hardware and the experiments show that, given sufficient training data, reasonable performance is achieved.

Crabs and lobsters are valuable crustaceans that contribute enormously to the seafood needs of the growing human population. This paper presents a comprehensive analysis of single- and multi-stage object detectors for the detection of crabs and lobsters using images captured onboard fishing boats. We investigate the speed and accuracy of multiple object detection techniques using a novel dataset, multiple backbone networks, various input sizes, and fine-tuned parameters. We extend our work to train lightweight models to accommodate the fishing boats equipped with low-power hardware systems. Firstly, we train Faster R-CNN, SSD, and YOLO with different backbones and tuning parameters. The models trained with higher input sizes resulted in lower frames per second (FPS) and vice versa. The base models were highly accurate but were compromised in computational and run-time costs. The lightweight models were adaptable to low-power hardware compared to the base models. Secondly, we improved the performance of YOLO (v3, v4, and tiny versions) using custom anchors generated by the k-means clustering approach using our novel dataset. The YOLO (v4 and it’s tiny version) achieved mean average precision (mAP) of 99.2% and 95.2%, respectively. The YOLOv4-tiny trained on the custom anchor-based dataset is capable of precisely detecting crabs and lobsters onboard fishing boats at 64 frames per second (FPS) on an NVidia GeForce RTX 3070 GPU. The Results obtained identified the strengths and weaknesses of each method towards a trade-off between speed and accuracy for detecting objects in input images.