Fish Counts Research Articles

Hybrid Swin-CSRNet: A Novel and Efficient Fish Counting Network in Aquaculture

Real-time estimation of fish biomass plays a crucial role in real-world fishery production, as it helps formulate feeding strategies and other management decisions. In this paper, a dense fish counting network called Swin-CSRNet is proposed. Specifically, the VGG16 layer in the front-end is replaced with the Swin transformer to extract image features more efficiently. Additionally, a squeeze-and-excitation (SE) module is introduced to enhance feature representation by dynamically adjusting the importance of each channel through “squeeze” and “excitation”, making the extracted features more focused and effective. Finally, a multi-scale fusion (MSF) module is added after the back-end to fully utilize the multi-scale feature information, enhancing the model’s ability to capture multi-scale details. The experiment demonstrates that Swin-CSRNet achieved excellent results with MAE, RMSE, and MAPE and a correlation coefficient R2 of 11.22, 15.32, 5.18%, and 0.954, respectively. Meanwhile, compared to the original network, the parameter size and computational complexity of Swin-CSRNet were reduced by 70.17% and 79.05%, respectively. Therefore, the proposed method not only counts the number of fish with higher speed and accuracy but also contributes to advancing the automation of aquaculture.

Automated fish counting system based on instance segmentation in aquaculture

Non-invasive methods of fish counting are crucial for the effective management of aquaculture and aquatic resources. However, accurate and reliable estimation of fish populations in real-time is a significant challenge due to severe occlusion, body deformation, and rapid movement of fish. Recent advancements in computer vision and deep learning algorithms offer substantial potential to revolutionize fish counting methodologies, thereby enhancing conservation efforts and promoting the sustainable management of aquatic resources. This paper presents a comprehensive dataset of 256,580 fish instances annotated with mask information at the pixel level. To address the challenges associated with geometric variations in fish size, pose, perspective, and body shape, we propose a lightweight instance segmentation model based on YOLOv8, which integrates the CSP bottleneck with dual convolutions and the deformable convolutional networks (DConv). In addition, an efficient feature fusion network incorporating multi-scale information was introduced to improve segmentation accuracy by enhancing the representation of feature information, and a decoupled head with attention mechanisms was designed to detect and segment fish bodies in real time. Our proposed approach achieved an outstanding mAP50 of 98.1 % at 60.6 FPS, with model parameters and floating point operations per second (FLOPS) amounting to 25.4 M and 96.9 G, respectively. The instance segmentation-based counting system exhibited impressive performance metrics, with Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R-Square (R2) values of 1.19, 1.65, and 0.992, respectively. The proposed method demonstrates strong potential for practical application as an intelligent solution for biomass estimation in recirculating aquaculture systems.

Fishing event detection and species classification using computer vision and artificial intelligence for electronic monitoring

Fisheries regulations require detailed catch reporting on commercial fishing vessels. Vital components for the sustainable management of fish stocks include a robust estimate of the number of fish caught and the species composition. Catch recording is often done manually by human observers on fishing vessels. Human observers are costly, and consistent data streams can be subject to observer availability and the weather. On-vessel cameras (electronic monitoring, EM) are a growing alternative to human observers. However, on-land human auditors are required to review hundreds of hours of videos recorded during fishing trips that can last for weeks. In this paper, a framework is presented to automatically detect fish in EM videos, count the total fishing events, and classify the fish species. For this purpose, a deep learning and computer vision-based model is developed to efficiently detect fish and fishers onboard a vessel. Secondly, a vision-based tracking pipeline tracks the detected fish and counts the total fishing events in the videos. Thirdly, the extracted fishing events are classified through a deep learning-based fish species classifier, to provide the distribution of different fish species caught for a fishing trip. For our experiments, the datasets were prepared using the electronic monitoring data of multiple fishing trips of a fishing vessel. The videos were recorded on Australian longline vessels targeting tunas and billfish. For the fish detection task, video frames were extracted and labelled manually to provide a digital ground-truth. For the fish species classification task, hundreds of fish images of multiple species were cropped to provide a training dataset for the fish classifier. For the fish counting task, manual counts for the fishing events of individual fish species were generated for the test fishing trips. The developed fish and fisher detector achieves a mean Average Precision of 87.0 % for fish and 94.0 % for fishers on test video frames. The fishing event detection pipeline achieves an Average Precision of 81.0 % and an Average Recall of 74.5 % on test videos. The fish species classifier achieves an Accuracy (Top-1) of 91.11 % for the classification of cropped fish images and 89.05 % for the classification of extracted fishing events from the videos. Experimental results show that our proposed computer vision and artificial intelligence-based solution for video analysis has great potential to automate the auditing process from electronic monitoring footage and contribute to the sustainable management of fish stocks.

A Fish Target Identification and Counting Method Based on DIDSON Sonar and YOLOv5 Model

In order to more accurately and quickly identify and count underwater fish targets, and to address the issues of excessive reliance on manual processes and low processing efficiency in the identification and counting of fish targets using sonar data, a method based on DIDSON and YOLOv5 for fish target identification and counting is proposed. This study is based on YOLOv5, which trains a recognition model by identifying fish targets in each frame of DIDSON images and uses the DeepSort algorithm to track and count fish targets. Field data collection was conducted at Chenhang Reservoir in Shanghai, and this method was used to process and verify the results. The accuracy of random sampling was 83.56%, and the average accuracy of survey line detection was 84.28%. Compared with the traditional method of using Echoview to process sonar data, the YOLOv5 based method replaces the step that requires manual participation, significantly reducing the time required for data processing while maintaining the same accuracy, providing faster and more effective technical support for monitoring and managing fish populations.

A method for counting fish based on improved YOLOv8

In industrial aquaculture, accurately counting fish in real-time is crucial for optimizing feeding strategies, preventing disease, and managing water quality. Current methods utilizing sensors, acoustics, machine learning, and density map regression face challenges such as high costs, invasiveness, and computational complexity. To address these limitations, we propose the YOLOv8n-MEMAGD method for accurate real-time fish counting. This method enhances YOLOv8 by incorporating the GELU activation function, MPDIoU for localization loss, and the C2f-FAM and C2f-MSCA modules into the backbone network. Additionally, the neck network is redesigned with a gather-and-distribution mechanism. Experimental results under land-based industrial aquaculture conditions demonstrate that YOLOv8n-MEMAGD achieved a mean absolute error (MAE) of 2.28 and a root mean square error (RMSE) of 2.84, even in challenging conditions such as fish overlap, aggregation, and background confounding. Compared to YOLOv8n, the proposed method increased average precision (AP50) for fish detection by 6.2 %, and significantly reduced MAE and RMSE by 61.4 % and 65.2 %, respectively, compared to CSRNet. Additionally, the method achieved a frame rate of 61 frames per second (FPS) when the number of fish ranged from 79 to 91, representing a 390.4 % increase over CSRNet. By comparing heatmaps, the proposed method demonstrates more effective detection of fish edge contours than current advanced algorithms. In conclusion, the proposed method shows promise for application in aquacultural scenarios with higher turbidity and larger number of fish.

Viability of high-frequency environmental DNA (eDNA) sampling as a fish enumeration tool

Despite growing interest in environmental DNA (eDNA) monitoring as a low-cost, non-invasive tool for monitoring fish presence and relative abundance, quantitative comparisons between fish abundance and eDNA are still rare, especially for anadromous species of conservation concern. To examine the relationship between eDNA data and management-relevant abundance data, we assessed eDNA abundance of river herring, an ecologically relevant anadromous sister-species pair, alongside sonar-generated fish counts at a well-characterized site in the Choptank River, Chesapeake Bay, MD. Over the 2021 spawning season, there was a strong correlation between daily eDNA concentration and daily counts of fish swimming upstream (Spearman’s rho = 0.86). A generalized additive model for location scale and shape (GAMLSS) was used to estimate fish count as a function of flow-corrected eDNA concentration and relevant covariates. Model- and sonar-generated run counts were 729,385 and 733,208 fish respectively, differing by just 0.5 %. A sub-sampling analysis showed that a model built on at least five weeks of data could utilize eDNA abundance to produce accurate estimates for the remainder of the season, while a reduction in eDNA sampling frequency substantially reduced run count accuracy. This work emphasizes the utility of quantitative eDNA monitoring for river herring populations and shows the necessity of high-frequency eDNA sampling to estimate run count.

Description of regression models for predicting the dynamics of pink salmon returns in the Kamchatka region based on climate-oceanological and population-genetic data

Several regression models for predicting returns of pink salmon in the Kamchatka region are presented. The data for 1990–2023 were analyzed. Among available climatic and oceanological indices, the most suitable for using as predictors for forecasting of pink salmon returns were the Pacific Decadal Oscillation (PDO) index, Western Pacific Cyclonic Index (WP), Arctic Oscillation (AO) index, and the sea surface temperature anomaly in the North Pacific. Multi-dimensional models of the «stock–recruitment» type were built on identified statistical patterns, which allowed to estimate potential abundance of the pink salmon returns to northeastern and western Kamchatka. Besides, methods for predicting the abundance of pink salmon returns on the data of fish counting in the sea are considered, using the materials of TINRO trawl surveys conducted in the Bering and Okhotsk Seas in the fall seasons of 2012–2023. To determine the abundance of pink salmon originated from West Kamchatka, genetic identification of regional composition of juveniles in mixed trawl catches was used. All tested methods have a high level of determination, but simpler regressive models are more prospective for practical forecasting of general trend in dynamics of pink salmon stocks in the Kamchatka region due to very weak generalization ability of more complicated models.

Multi-species identification and number counting of fish passing through fishway at hydropower stations with LigTraNet

Fishway monitoring can verify the effectiveness of the fishway, optimise the operation mode, and achieve scientific management of fishway operations. Traditional fishway monitoring approaches, hindered by their inefficiency and substantial disruption of fish, are ill-suited for long-term surveillance; thus, employing video monitoring coupled with object detection technology presents an alternative or complementary solution. However, challenges such as the constrained computational capacity of onsite equipment in fishways, complexities involved in model deployment, and sluggish pace of detection are significant hurdles. In this study, by utilising the YOLOv8n model as a benchmark, we engineered a cross-stage partial module with a single convolution (C1) module to replace the existing C2f module with the aim of enhancing performance. We replaced the conventional 2D convolutions in the bottleneck configuration with depthwise separable convolutions and integrated the SimAM module to extract the detailed characteristics of the fish species. By amalgamating LigObNet detection with the DeepSORT algorithm, we established LigTraNet, which is designed to enable precise tracking, identification, and counting of individual fish. The results showed that LigObNet exhibited the lowest complexity and fastest detection speed for underwater fish among similar object recognition and detection models. Compared with the benchmark YOLOv8n model, there were reductions of 8.9% in the network layers, 40.5% in the parameter count, 39.3% in the memory footprint, and 35.8% in the giga floating-point operations and a 38.1% improvement in the inference speed. LigTraNet achieved a total count accuracy rate of 91.8%, demonstrating superior species quantification capabilities over other models with minimal resource usage and rapid inference capabilities, thus offering enhanced practicality for deployment on devices in real-world engineering contexts. This represents a departure from traditional manual monitoring methods for assessing fishway effectiveness, revolutionising aquatic ecological monitoring tools and methodologies and fostering the collaborative advancement of water resource project operations and ecological conservation.

Optimal level of dietary arginine enhances growth performance, haemato-biochemical status, metabolic responses and growth-related gene expression of GIFT juveniles reared in inland saline water

The effects of arginine levels in diets on growth performance, survival, and haemato-biochemical responses and gene expression of genetically improved farmed tilapia (GIFT) juveniles raised in inland saline water (ISW) of 10 ppt were evaluated through a 60-day growth trial. Seven isoproteinaceous (⁓320 g protein/kg), isolipidic (80 g/kg) and isoenergetic (16.8 MJ DE/kg) diets with graded levels of arginine viz. A10 (10 g/kg), A12.5 (12.5 g/kg), A15 (15.0 g/kg), A17.5 (17.5 g/kg), A20 (20.0 g/kg), A22.5 (22.5 g/kg) and A25 (25.0 g/kg) were prepared. Acclimated GIFT juveniles (3.75 ± 0.017 g) were randomly assigned into seven treatments in triplicate with 15 fish/replicate and fed with respective diets thrice a day on a satiation basis. The percent weight gain, specific growth rate, protein efficiency ratio, whole carcass protein, tissue alanine and aspartic acid aminotransferase activities, serum total protein, albumin and globulin concentration and expression of IGF-1 & IGF-1R genes of fish significantly increased, while feed conversion ratio significantly decreased with increasing arginine in the diet up to 17.5 g/kg and decreased thereafter. Additionally, white blood cell count increased with rising arginine levels in the diet. However, body indices, survival, whole carcass lipid, total ash and essential amino acids contents, activities of digestive and antioxidant enzymes, serum glucose level and blood haemoglobin level and red blood cell count of fish remained unchanged for different treatments. Based on broken-line regression and second order polynomial regression models in relation to hepatic IGF-1 gene expression and percent weight gain, the dietary optimal arginine level was found to be 15.97 and 16.76 g/kg and 17.06 and 17.41 g/kg, respectively with the optimum range of 15.97 to 17.41 g/kg for successful culture of GIFT in ISW of 10 ppt salinity.

Adaptive density guided network with CNN and Transformer for underwater fish counting

Accurate assessment of high-density underwater fish resources is vital to the aquaculture industry. It is directly related to the formulation of fishery insurance strategies and the implementation of breeding plans. However, accurately counting fish in high-density environments becomes challenging due to the uneven distribution of fish density and individual fish’s different sizes and postures. To break through this technical bottleneck, we developed an advanced adaptive density-guided high-density fish counting network. In detail, first of all, the network adopts a multi-layer feature fusion structure similar to UNet, which significantly enhances the matching between fish targets of different scales and feature pyramid levels, effectively alleviating the problems caused by scale changes and morphological deformations. Secondly, the network also introduces a density-guided adaptive selection module, which can intelligently judge the applicability of Convolutional Neural Network and Transformer blocks in different density areas, thereby achieving robust information transfer and interaction between blocks. Finally, to verify the effectiveness of this method, we also specially constructed two high-density data sets: a simulated high-density underwater fish image data set (SHUFD) and a real high-density underwater fish image data set (RHUFD). The proposed method has significant improvements over the state-of-the-art method (CUT) on SHUFD and RHUFD datasets, with the mean absolute error, mean square error, background region bias, foreground region bias and density map bias indicators improving by 3.44% and 6.47%, 11.43% and 4.41%, 23.91% and 29.48%, 4.43% and 10.33%, 8.3% and 13.14%, respectively.

Sonar Fish School Detection and Counting Method Based on Improved YOLOv8 and BoT-SORT

Fish object detection and counting in pelagic fisheries face many challenges in complex environments. Sonar imaging technology offers a solution because it generates high-resolution images underwater. In this paper, we propose a sonar-based fish object detection and counting method using an improved YOLOv8 combined with BoT-SORT to address issues such as missed detection, false detection, and low accuracy caused by complex factors such as equipment motion, light changes, and background noise in pelagic environments. The algorithm utilizes the techniques of lightweight upsampling operator CARAFE, generalized feature pyramid network GFPN, and partial convolution. It integrates with the BoT-SORT tracking algorithm to propose a new region detection method that detects and tracks the schools of fish, providing stable real-time fish counts in the designated area. The experimental results indicate that while focusing on maintaining a lightweight design, the improved algorithm achieved a 3.8% increase in recall and a 2.4% increase in mAP0.5 compared to the original algorithm. This significantly impacts scientific and rational fishery planning, marine resource protection, and improved productivity. At the same time, it provides important data support for marine ecological monitoring, environmental protection, and fishery management, contributing to sustainable fishery development and marine ecology preservation.

Effect of Salt (Nacl) and Vacuum Packaging on Some of the Quality Attributes of Snakehead Fish

Snakehead fish (Chana striata) is one of the important protein sources among Indonesian people. However, the shelf life is short leading to limited use of fresh snakehead fish. This study aimed to analyse the effect of vacuum packaging and addition of salt on the sensory properties, total plate count, water activity and pH of fresh snakehead fish. The fish was salted (NaCl) at various concentrations, including 0, 5, 10 and 15%, and kept for 14 days at refrigeration temperature (± 3°C). Sensory evaluation, total plate count (TPC) and water activity (aw) measurements, and pH analysis were performed at day 0, 7 and 14. All groups with salt addition exhibited a significantly slower decrease in the mean scores (9-point hedonic scale) of color, aroma, appearance and texture (p < 0.05), with the decrease for all sensory properties detected at day 7 of storage. In microbial tests, a significant increase in TPC was observed in the groups with 0% and 5% salt addition (p < 0.05). The group with 15% salt addition showed the lowest water activity. Also, the group without salt addition had the highest pH value. This study demonstrated that addition of salt to snakehead fish might gives beneficial effects on the shelf life of a vacuum packaged product. However, the optimum shelf life using vacuum packaging and salt addition needs further study.

Advances in the application of stereo vision in aquaculture with emphasis on fish: A review

AbstractThe effective implementation of machine vision has played a crucial role in advancing intelligent aquaculture across various domains. Stereo vision, as a branch of machine vision, has become a mainstream technology in aquaculture. Its distinctive capability to conduct comprehensive underwater monitoring from multiple angles, unaffected by object occlusion has propelled it to the forefront of aquaculture applications. This article offers a comprehensive review of the diverse applications of stereo vision in aquaculture spanning from its inception to present. The exploration encompasses its role in crucial areas such as biomass estimation and behavioural analysis, which include fish counting, weight estimation, swimming behaviour, feeding behaviour and abnormal behaviour. Furthermore, the paper delves into the advantages of stereo vision over traditional 2D machine vision approaches, while also acknowledging limitations, and identifying future challenges that must be addressed to fully leverage its potential in aquaculture. The review emphasizes the prospect of advancement in deep learning stereo‐matching algorithms specifically designed for underwater environments to catalyse a breakthrough in stereo vision technology. In summary, this review aims to provide researchers and practitioners with a better understanding of the current development of stereo vision in aquaculture, optimizing stereo vision technology and better serving the aquaculture field.

Occurrence of Antimicrobial-Resistant Bacteria in Intestinal Contents of Wild Marine Fish in Chile.

Antimicrobial-resistant bacteria (ARB) from the intestinal contents of wild fish may have a relevant ecological significance and could be used as indicators of antimicrobial-resistance dissemination in natural bacterial populations in water bodies impacted by urban contamination. Thus, the occurrence of ARB in the intestinal contents of pelagic and demersal wild fishes captured in anthropogenic-impacted Coquimbo Bay in Chile was studied. Culturable counts of total and antimicrobial-resistant bacteria were determined by a spread plate method using Trypticase soy agar and R2A media, both alone and supplemented with the antimicrobials amoxicillin, streptomycin, florfenicol, oxytetracycline and ciprofloxacin, respectively. Heterotrophic plate counts of pelagic and demersal fishes ranged from 1.72 × 106 CFU g-1 to 3.62 × 109 CFU g-1, showing variable proportions of antimicrobial resistance. Representative antimicrobial-resistant isolates were identified by 16S rRNA gene sequencing, and isolates (74) from pelagic fishes mainly belonged to Pseudomonas (50.0%) and Shewanella (17.6%) genera, whereas isolates (68) from demersal fishes mainly belonged to Vibrio (33.8%) and Pseudomonas (26.5%) genera. Antimicrobial-resistant isolates were tested for susceptibility to 12 antimicrobials by an agar disk diffusion method, showing highest resistance to streptomycin (85.2%) and amoxicillin (64.8%), and lowest resistance to oxytetracycline (23.2%) and ciprofloxacin (0.7%). Only furazolidone and trimethoprim/sulfamethoxazole were statistically different (p < 0.05) in comparisons between isolates from pelagic and demersal wild fishes. Furthermore, an important number of these isolates carried plasmids (53.5%) and produced Extended-Spectrum-β-lactamases (ESBL) (16.9%), whereas the detection of Metallo-β-Lactamases and class 1-integron was rare. This study provides evidence that wild fish are important reservoirs and spreading-vehicles of ARB, carrying plasmids and producing ESBLs in Chilean marine environments.

Real-Time Smart Fish Feeder Model Using Ssd Mobile Net V2

The success of fish farming will depend on improved feed management and lower operating costs, which are essential factors in facilitating an efficient food allocation to the fish. Various automatic fish feeders were used to feed the fish at set intervals, it consists of a mechanical and electrical system to form a device and execute a programmed method, instead of manually feeding the fish by hand. Many methods are effective at evaluating and quantifying fish feeding intensity but are mostly done on the movement and behavior of the fish. However, recognition accuracy is affected due to water quality and the overlapping of fish. To solve this problem, in this study, the author will be focusing on the model in counting the fish pellets, it will capture the image, and count the pellets and the results will be a novel method as a basis for releasing pellets in laying the logical foundation in creating a modern real-time smart fish feeder. The model produced a significant result that detected small objects like fish pellet and count that has gathered a minimal loss in terms of classification, localization, regularization, and normalization.

The development and usability test of an automated fish counting system based on CNN and contrast limited histogram equalization

The aquaculture industry has rapidly grown over the year. One pertinent aspect is the ability of the aquaculture farm management to accurately count the fish populations to provide effective feeding and the control of breeding density. The current practice of counting the fish manually increased the hatchery workers workload and led to inefficiency. The presented work proposed an intelligent, web-based fish counting system to assist hatchery workers in counting fish from images. The methodology consists of two phases. First, an intelligent fish counting engine is developed. The captured image was first enhanced using the contrast limited adaptive histogram equalization. A deep learning architecture in the form of you only look once (YOLO)v5 is used to generate a model to identify and count fish on the image. Second, a web-based application is developed to implement the developed fish counting engine. When applied to the test data, the developed engine recorded a precision of 98.7% and a recall of 65.5%. The system is also evaluated by hatchery workers in the University Malaysia Sabah fish hatchery. The results of the usability and functionality evaluations indicate that the system is acceptable, with some future work suggested based on the feedback received.

Comparison of environmental DNA and underwater visual count surveys for detecting juvenile Coho Salmon in small rivers

AbstractObjectiveThis study compares the probability of detecting juvenile Coho Salmon Oncorhynchus kisutch using both environmental DNA (eDNA) techniques and underwater visual count (UVC) surveys in northern California rivers. Here, UVC surveys commonly have detection probabilities (p) surpassing 0.90, providing an ideal setting to examine the performance of newer eDNA methods. We also evaluate the potential for using eDNA concentrations to predict the count of Coho Salmon within pool habitats.MethodsWe conducted paired eDNA and UVC surveys in 96 pools across 25 stream reaches within the Smith River basin, California. Method‐specific p and the effect of environmental covariates were estimated using multiscale occupancy modeling. We used generalized linear models to evaluate the relationship of fish counts to eDNA concentrations and habitat covariates.ResultThe eDNA and UVC methods showed a high degree of agreement in detecting the presence of Coho Salmon within a pool (93% agreement) and survey reach (80% agreement). Detection probabilities for eDNA (peDNA) and for UVC (pUVC) were similar and high at median levels of pool residual depth and contributing basin area (peDNA = 91%, pUVC = 89%). Contributing basin area (a proxy for discharge) had a strong, negative effect that was more pronounced for peDNA than for pUVC (e.g., in the largest basins, peDNA = 34% whereas pUVC = 77%). We did not find eDNA concentrations to be a good predictor of Coho Salmon counts in small pools.ConclusionThis study demonstrates that eDNA methods yielded nearly identical results to UVC surveys in catchments &lt;36 km2 and can provide a highly effective approach for determining the distribution of Coho Salmon. However, additional investigation is required before eDNA could be used to estimate relative abundance in small pools.

FCFormer: fish density estimation and counting in recirculating aquaculture system

In intelligent feeding recirculating aquaculture system, accurately estimating fish population and density is pivotal for management practices and survival rate assessments. However, challenges arise due to mutual occlusion among fish, rapid movement, and complex breeding environments. Traditional object detection methods based on convolutional neural networks (CNN) often fall short in fully addressing the detection demands for fish schools, especially for distant and small targets. In this regard, we introduce a detection framework dubbed FCFormer (Fish Count Transformer). Specifically, the Twins-SVT backbone network is employed first to extract global features of fish schools. To further enhance feature extraction, especially in the fusion of features at different levels, a Bi-FPN aggregation network model with a CAM Count module is incorporated (BiCC). The CAM module aids in focusing more on critical region features, thus rendering feature fusion more cohesive and effective. Furthermore, to precisely predict density maps and elevate the accuracy of fish counting, we devised an adaptive feature fusion regression head: CRMHead. This approach not only optimizes the feature fusion process but also ensures superior counting precision. Experimental results shown that the proposed FCFormer network achieves an accuracy of 97.06%, with a mean absolute error (MAE) of 6.37 and a root mean square error (MSE) of 8.69. Compared to the Twins transformer, there's a 2.02% improvement, outperforming other transformer-based architectures like CCTrans and DM_Count. The presented FCFormer algorithm can be effectively applied to fish density detection in intelligent feeding recirculating aquaculture system, offering valuable input for the development of intelligent breeding management systems.

Effects of installation depth and recording start time on fish count and mean fork length of &lt;i&gt;Seriola quinqueradiata&lt;/i&gt; measured by optical stereo cameras

Effects of installation depth and recording start time on fish count and mean fork length of &lt;i&gt;Seriola quinqueradiata&lt;/i&gt; measured by optical stereo cameras

A video monitoring and computational system for estimating migratory juvenile fish abundance in river systems

AbstractDiadromous fishes migrate between marine and fresh waters for reproduction. For anadromous species, which spawn in freshwater, improved access to freshwater spawning and nursery habitats and ability of juveniles to emigrate to the ocean may support population recovery. Despite the potentially enormous influence of early life stage survival on adult population size, managers and scientists have limited capacity to assess numbers of juvenile anadromous fishes leaving freshwater ecosystems. Such data are critical for evaluating reproductive success and habitat suitability and have been identified as a top priority in anadromous fish research and management. We developed a state‐of‐the‐art underwater video and computational system to collect videos to estimate abundances and migration timing for juvenile river herring (Alosa pseudoharengus; Alosa aestivalis). We collected continuous video in the Monument River (Bourne, Massachusetts, USA) from June to November 2017. We trained three types of neural network models to detect and count fish in video frames and evaluated model performance by comparing human counts to model outputs. Our top model assessed presence and absence (F1 = 87%) and counted fish (counting error 9.4%) with an accuracy comparable to human counters (F1 = 88%). Our system's capability to collect accurate counts of emigrating juveniles will provide critical information that could be related to the numbers of spawning adults, system‐specific productivity, and spawning and nursery habitat suitability. Both the video collection system and computational model may be transferrable to other sites and for other species where tracking juvenile emigration may inform management efforts.