Fish Classification Research Articles

Improving CNN Fish Detection and Classification with Tracking

The regular and consistent monitoring of marine ecosystems and fish communities is becoming more and more crucial due to increasing human pressures. To this end, underwater camera technology has become a major tool to collect an important amount of marine data. As the size of the data collected outgrew the ability to process it, new means of automatic processing have been explored. Convolutional neural networks (CNNs) have been the most popular method for automatic underwater video analysis for the last few years. However, such algorithms are rather image-based and do not exploit the potential of video data. In this paper, we propose a method of coupling video tracking and CNN image analysis to perform a robust and accurate fish classification on deep sea videos and improve automatic classification accuracy. Our method fused CNNs and tracking methods, allowing us to detect 12% more individuals compared to CNN alone.

REAL-TIME DETECTION AND CLASSIFICATION OF FISH IN UNDERWATER ENVIRONMENT USING YOLOV5: A COMPARATIVE STUDY OF DEEP LEARNING ARCHITECTURES

This article explores techniques for the detection and classification of fish as an integral part of underwater environmental monitoring systems. Employing an innovative approach, the study focuses on developing real-time methods for high-precision fish detection and classification. The implementation of cutting-edge technologies, such as YOLO (You Only Look Once) V5, forms the basis for an efficient and responsive system. The study also evaluates various approaches in the context of deep learning to compare the performance and accuracy of fish detection and classification. The results of this research are expected to contribute to the development of more advanced and effective aquatic monitoring systems for understanding underwater ecosystems and conservation efforts.

Checklist of marine fishes in the Beibu Gulf: fish classification, resource protection, and biodiversity challenge

Checklist of marine fishes in the Beibu Gulf: fish classification, resource protection, and biodiversity challenge

Comparison and phylogenetic analysis of the mitochondrial genomes of Synodontis eupterus and Synodontis polli

We aimed to distinguish Synodontis eupterus and Synodontis polli. We performed sequencing and bioinformatic analysis of their mitochondrial genomes and constructed a phylogenetic tree of Mochokidae fish using maximum likelihood and Bayesian methods based on protein-coding gene (PCG) sequences of 14 Mochokidae species. The total length of the S. eupterus mitochondrial genome was 16,579 bp, including 13 (PCGs), 22 tRNA genes, two rRNA genes, and one D-loop, with an AT-biased nucleotide composition (56.0%). The total length of the S. polli mitochondrial genome was 16,544 bp, including 13 PCGs, 22 tRNA genes, two rRNA genes, and one D-loop, with an AT-biased nucleotide composition (55.0%). In both species, except for COI, PCGs use ATG as the starting codon, the vast majority use TAG or TAA as the ending codon, and a few use incomplete codons (T - or TA -) as the ending codon. Phylogenetic analysis showed that S. eupterus and Synodontis clarias converged into one branch, S. polli and Synodontis petricola converged into one branch, Mochokiella paynei, Mochokus brevis, and nine species of the genus Synodontis converged into one branch, and M. paynei clustered with the genus Synodontis. This study lays a foundation for rebuilding a clearer Mochokidae fish classification system.

Classification of Marine Fish Based on Image Using Convolutional Neural Network Algorithm and VGG16

Marine fish are one of the many natural resources that frequently consumed by people. However, several types of marine fish are prohibited from consumption due to being nearly extinct. Additionally, some fish species contain high levels of mercury that can be harmful to human if consumed. Because of the large number of marine fish species, it becomes challenging to identify them without knowledge of fisheries. Moreover, computers have become highly advanced devices that facilitate various human activities. This advancement allows for the creation of systems capable of processing information from images, known as image classification. There are numerous methods that can be employed in designing an image classification system, one of which is transfer learning. This study was aimed to design an image classification system using the transfer learning method with a pre-trained VGG16 model and Convolutional Neural Network algorithm. The research results showed a training data accuracy of 100% and a validation data accuracy of 99.3%, with an overall accuracy of 84% and a loss value of 0.6591.

ConvFishNet: An efficient backbone for fish classification from composited underwater images

For the purpose of monitoring fish health, managing aquaculture, and comprehending marine ecology, there is a growing interest in the automatic classification of different fish species. Recent developments in machine vision-based classification methods, known for their speed and non-destructive nature, have led to the introduction of various automatic categorization approaches. Drawing inspiration from FishNet, a new architecture model named ConvFishNet has been proposed. This model incorporates large convolutional kernels and depth-wise separable convolutions to reduce the number of parameters in the model. Additionally, the PixelShuffle has been developed to enhance the upsampling information and improve fish classification performance. While maintaining precision in fish classification, the model achieves a lightweight design with only 0.83G. Compared with the FishNet model, the new model reduces parameters by 80 %. The model demonstrates a precision of 88.44 % on the WildFish dataset subset and 99.8 % on the Fish4knowledge dataset, surpassing existing methods including FishNet. This method shows promise for fish classification in challenging underwater environments, such as marine and aquaculture settings, and further investigation is planned for the future.

Identifying losers: Automatic identification of growth-stunted salmon in aquaculture using computer vision

During the production of salmonids in aquaculture, it is common to observe growth-stunted individuals. The cause for the so-called “loser fish syndrome” is unclear, which needs further investigation. Here, we present and compare computer vision systems for the automatic detection and classification of loser fish in Atlantic salmon images taken in sea cages. We evaluated two end-to-end approaches (combined detection and classification) based on YoloV5 and YoloV7, and a two-stage approach based on transfer learning for detection and an ensemble of classifiers (e.g., linear perception, Adaline, C-support vector, K-nearest neighbours, and multi-layer perceptron) for classification. To our knowledge, the use of an ensemble of classifiers, considering consolidated classifiers proposed in the literature, has not been applied to this problem before. Classification entailed the assigning of every fish to a healthy and a loser class. The results of the automatic classification were compared to the reliability of human classification. The best-performing computer vision approach was based on YoloV7, which reached a precision score of 86.30%, a recall score of 71.75%, and an F1 score of 78.35%. YoloV5 presented a precision of 79.7%, while the two-stage approach reached a precision of 66.05%. Human classification had a substantial agreement strength (Fleiss’ Kappa score of 0.68), highlighting that evaluation by a human is subjective. Our proposed automatic detection and classification system will enable farmers and researchers to follow the abundance of losers throughout the production period. We provide our dataset of annotated salmon images for further research.

Walleye (Sander vitreus, Mitchill 1818) age and sex classification using innovative supervised and unsupervised machine learning and soft computing methodologies

Walleye (Sander vitreus) is a freshwater perciform native to Northern America and Canada with high commercial value. Stocking programs for the species are unable to supply the significant demand, creating stock management problems and an ecosystemic decline. For sustainable fisheries, the age and sex estimation are fundamental to marine stock management. Biological methods for age determination and sex estimation were used in the past, conventional observations and techniques are substituted with machine learning methods, currently underused. The aim of this study is to evaluate the performance of an ensemble of Machine Learning (ML) methods for age and sex classification of walleye. Based on the performance of the ensemble models, appropriate evaluation decides on the age criterion with near optimal performance using supervised and unsupervised ML algorithms on the given fish datasets. The innovation of this work lies in the fact that the species age or sex estimation is based on a paramount involvement of models that diminish the proprietary idiosyncrasies of individual methods on specific datasets. For the same models two approaches were used: applied model algorithms developed both in R and Python (Orange Rapid Application Development environment) programming language provided high accuracy percentages. Among all methodologies, the Decision Tree method provided the best classification, for age estimation with 96% and 98% accuracy using the R and Orange models respectively. For the sex classification, Random Forest performed the best using the R-model with 94% accuracy, outperformed by the AdaBoost model in Orange with 97.6% accuracy. When compared, the models which were developed in Python generally performed better than the corresponding R-models both in age and sex classification. Overall, the proposed ML approach is emphasized as a rapid economic solution to standard fish classification with high accuracy, and thus could be integrated in stock assessment for a more sustainable aquaculture management.

Betta fish classification using transfer learning and fine-tuning of CNN models

Betta fish, known as freshwater fighters, are in demand because of their beauty and characteristics. These betta fish such as Crowntail, Halfmoon, Doubletail, Spadetail, Plakat, Veiltail, Paradise, and Rosetail are hard to recognize without knowledge about them. Therefore, transfer learning of Convolutional Neural Network models was proposed to classify the betta fish from the image. The transfer learning process used a pre-trained model from ImageNet of VGG16, MobileNet, and InceptionV3 and fine-tuned the models on the betta fish dataset. The models were trained on 461 images, validated with 154 images, and tested on 156 images. The result shows that the InceptionV3 model excels with 0.94 accuracies compared to VGG16 and MobileNet which acquire 0.93 and 0.92 accuracy respectively. With good accuracy, the trained model can be used in betta fish recognition applications to help people easily identify betta fish from the image.

Creation of Machine Learning-Based Fish Classification Systems Based on Morphometric and Mathematical Transform Data

The accurate classification of fish species is crucial for various ecological, conservation, and fisheries management applications. Traditional methods relying solely on manual observation or morphological characteristics can be time-consuming and prone to errors. In recent years, advancements in machine learning (ML) techniques coupled with morphometric and mathematical transform data have shown promising results in automating and enhancing fish species classification processes. This paper presents a comprehensive review of the development and application of ML-based fish classification systems utilizing morphometric measurements and mathematical transforms. The core of this paper focuses on the integration of morphometric and mathematical transform data with ML algorithms for fish classification. In this review commonly employed ML algorithms, including but not limited to support vector machines (SVM), artificial neural networks (ANN), random forests (RF), and convolutional neural networks (CNN). Furthermore, this work discuss feature selection techniques to optimize classification performance and reduce dimensionality. These may include the incorporation of multimodal data sources, transfer learning approaches, and the development of user-friendly tools for field biologists and conservationists. Overall, this paper serves as a comprehensive guide for researchers and practitioners interested in leveraging ML techniques for accurate and efficient fish species classification.

Shrimp and Fish Underwater Image Clasification Using Features Extraction and Machine Learning

Shrimp cultivation is one type of cultivation that has a significant impact on the social status of coastal communities. Shrimp farming traditionally faces several challenges, including water pollution, imbalances in temperature, feed, media, and costs. Monitoring the condition of shrimp in the cultivation environment is very necessary to determine the condition of shrimp in the water. Classification of shrimp and fish is the first step in monitoring the condition of shrimp underwater. This research proposes the development of a method for classifying shrimp and fish underwater using feature extraction and machine learning. The flow of this research is: (1) preparing data from ROI detection results, (2) extraction process of morphometric characteristics P and T, (3) calculating the value of morphometric characteristics P and T, (4) data breakdown for training data and testing data, (5) Model creation process, data training and data testing using SVM, RF, DT, and KNN, (6) Evaluation of classification results using a confusion matrix. From this research, it was found that the Random Forest method obtained the highest accuracy, namely 0.93. From this matrix, the values ​​obtained are True Positive = 349, False Positive = 28, True Negative = 223, False Negative = 0.

MLP Neural Networks in Sex Classification of a High Commercial Fish on the Artisanal Fisheries in Brazil

The scarcity of data on artisanal fishing, particularly in tropical South American countries, presents an additional hurdle for fisheries management and marine conservation. While artificial intelligence (AI) has found applications in various domains, including marine sciences, most AI models primarily focus on species identification. Unfortunately, these models often overlook data-limited scenarios. This article examines the potential of an MLPtype ANN model in addressing this gap. The model aims to classify the sex of a dioecious (separate sexes) fish species of high commercial importance, shedding light on its implications for fisheries management. Evaluation of the model's classification performance using precision, recall, f1-score, and accuracy reveals promising results exceeding 80% for both sexes across both training and testing phases. These findings underscore the potential of MLP models in aiding Brazil's fishing sector management in grappling with challenges stemming from data scarcity. By providing efficient information essential for decision-making regarding the management of specific fishing stocks, such models offer valuable insights into effective fisheries management strategies.

Hydro-acoustic classification and abundance estimation of mesopelagic fish in deep scattering layers (DSL) of the Indian Ocean

Abstract Deep scattering layers (DSL) in oligotrophic systems are typically comprised of a variety of coexisting organisms, including fish, zooplankton, jellyfish, and squid, and hence there is uncertainty about the proportion of the acoustic backscatter in these layers that can be attributed to mesopelagic fish. Here, acoustic targets were classified using a multi-frequency acoustic classification algorithm based on data collected in the Indian Ocean at 18, 38, 70, and 120 kHz during three acoustic surveys in 2018. Frequency-dependent backscattering strength information (∆Sv) was extracted from acoustic data that coincided with trawl hauls dominated by mesopelagic fish. Five ∆Sv frequency pairs were used to separate the acoustic backscattering into three broad scattering categories, i.e. mesopelagic fish, crustacean-like/tunicates, and squids/others. Results indicated that the DSL is highly diverse and dominated by mesopelagic fish, with average densities at a regional scale ranging from 12.0 (±10.9) to 26.0 (±21.7) g m−2, and proportions of the acoustic backscatter attributed to mesopelagic fish ranging from 0.5 to 0.7. These estimates are generally lower than previously estimated for other regions of the Indian Ocean. The situation may well be similar elsewhere, particularly in oligotrophic systems, with potential ramifications for global mesopelagic fish biomass estimates.

CMAF: Cross-Modal Augmentation via Fusion for Underwater Acoustic Image Recognition

Underwater image recognition is crucial for underwater detection applications. Fish classification has been one of the emerging research areas in recent years. Existing image classification models usually classify data collected from terrestrial environments. However, existing image classification models trained with terrestrial data are unsuitable for underwater images, as identifying underwater data is challenging due to their incomplete and noisy features. To address this, we propose a cross-modal augmentation via fusion ( CMAF ) framework for acoustic-based fish image classification. Our approach involves separating the process into two branches: visual modality and sonar signal modality, where the latter provides a complementary character feature. We augment the visual modality, design an attention-based fusion module, and adopt a masking-based training strategy with a mask-based focal loss to improve the learning of local features and address the class imbalance problem. Our proposed method outperforms the state-of-the-art methods. Our source code is available at https://github.com/WilkinsYang/CMAF .

Fine-Grained Fish Classification From Small to Large Datasets With Vision Transformers

Fine-Grained Fish Classification From Small to Large Datasets With Vision Transformers

Leveraging VGG16 for Fish Classification in a Large-Scale Dataset

When the VGG16 model was applied to fish picture classification, the overall accuracy was a remarkable 99%, demonstrating strong performance over most of the dataset. Still, a thorough assessment of the model's efficacy necessitates a look beyond its general accuracy. A more detailed evaluation is possible thanks to class-specific metrics like precision, recall, and F1-score, which provide information on how well the model performs on particular classes. Although the high overall accuracy is encouraging, more research into these metrics and the possibility of class imbalances should be taken into account to guarantee consistent performance in the fish image classification challenge across all categories. A more comprehensive assessment of the model's effectiveness benefits from a contextual knowledge of the application domain and a careful examination of evaluation measures.

Comparative Analysis VGG16 Vs MobileNet Performance for Fish Identification

This research aims to conduct a comparative evaluation of the efficacy of two neural network architectures in the field of fish identification through the utilization of supervised learning techniques. The evaluation of VGG16 and MobileNet, which are prominent deep learning architectures, has been conducted about their speed, accuracy, and efficiency in resource utilization. To assess the classification performance of both architectures, we employed a dataset encompassing diverse fish categories. The findings indicated that the VGG16 model demonstrated superior accuracy in fish classification, albeit due to increased computational time and resource utilization. On the contrary, MobileNet exhibits enhanced speed and efficiency, albeit at a marginal cost to its accuracy. The findings of this study have the potential to inform the selection of deep learning models for fish recognition scenarios, considering the specific requirements of the task, such as prioritizing accuracy or efficiency. The findings mentioned above offer significant insights that can be utilized in the advancement of Artificial Intelligence (AI)-based applications within the domains of fisheries resource management and environmental monitoring. These applications specifically necessitate precise and effective fish recognition capabilities. The comparison findings indicate that the accuracy achieved by VGG16 was 0.99, whereas MobileNet also attained an accuracy of 0.99.

Automatic detection and counting of fisheries using fish images

In Senegal, stock recovery and fish classification are based on manual data collection, and the fish caught by the fishery are not often declared. What's more, data collection suffers from a lack of tools for monitoring and counting fish caught at fishing docks. Researchers have carried out studies on the fishery in Senegal, but data collection is almost non-existent. Moreover, there is no local fisheries database or automatic detection and counting algorithm. In this paper, a semantic segmentation algorithm is proposed using intelligent systems for the collection of fishery catches, for the formation of the local database. The data are collected by taking images of fish at the Soumbédioune fishing wharf in Senegal, and are completed with the Fishbase database. They were applied to the algorithm and resulted in a segmented dataset with masks. This constitutes our local database. The database is used with YOLO v8. The latter is very important for detecting images with bounding boxes in order to train the model. The results obtained are very promising for the proposed automatic poison detection and counting model. For example, the recall-confidence scores translate into bounding box performance with scores ranging from 0.01 to 0.75, confirming the performance of this model with bounding boxes

Assessment of Morphometric Variations Among the Populations of Asian Sheat Catfish <i>Wallago attu</i> (Siluridae) from Five Indian Rivers

Morphological variations of the Asian sheat catfish, Wallago attu (Bloch and Schneider, 1801) sampled from the rivers Gomti, Ganga, Yamuna at Lucknow, Kanpur and Agra districts in Northern India, river Hooghly at Kolkata in Eastern India and river Pampa at Kerala in Southern India were analyzed in the present study. Images of 261 specimens of W. attu were taken without harming the fish to evaluate the morphometric variation using 8 digitized homologous landmarks for truss analysis and geometric morphometrics. Size corrected morphometric variables were analyzed using analysis of variance (ANOVA), multivariate analysis of variance (MANOVA), principal component analysis (PCA) and discriminant function analysis (DFA). Coefficients of all PCs were positively and negatively correlated to 31 significant (p 0.001) truss measurements due to variations in size and shape. Morphometric measurements lying in the head region such as head length (HL), eye diameter (ED), inter-orbital length (IOL) and snout length (SNL) contributed most to differentiate the populations in truss analysis. The ED, HL, SNL and IOL of Yamuna fish was significantly (p 0.001) smaller than the other fish populations. This could be attributed to the heavy pollution load in the Yamuna River and differences in the feeding regimes of fish. PCs and relative warps depicted differences in geometric analysis because of body shape variations among different river populations. The scatter plot in DFA revealed the presence of three distant and different phenotypic groups of W. attu among five wild river populations of fish with 70.5 and 83.9% of correct classification of fish into their original groups in truss and geometric analysis respectively. A single homogenous stock of fish was found to be in the rivers Gomti, Ganga and Hooghly because of a high level of intermixing (maximum misclassification) among the three river populations, whereas Yamuna and Pampa populations were two non-overlapping isolated groups due to negligible intermixing. These variations in the present study might be the effect of geographical isolation and river fragmentation due to the change in habitat that restricts fish movements. This is the first report on the morphometric assessment of W. attu that provides base line data for further confirmatory stock studies that would be useful for conservation and sustainable management of this vulnerable fish.

Attention ConvMixer Model and Application for Fish Species Classification

Exploring the ocean has always been one of the foremost challenges for humankind, and fish classification is one of the crucial tasks in this endeavor. Manual fish classification methods, although accurate, consume significant time, money, and effort, while computer-based methods such as image processing and traditional machine learning often fall short of achieving high accuracy. Recently, deep convolutional neural networks have demonstrated their capability to ensure both time efficiency and accuracy in this task. However, deep convolutional networks typically have a large number of parameters, requiring substantial training time, and the convolutional operations lack attentional mechanisms. Therefore, in this paper, we propose the AttentionConvMixer neural network with Priority Channel Attention (PCA) and Priority Spatial Attention (PSA). The proposed approach exhibits good performance across all three fish classification datasets without introducing any additional parameters, thus demonstrating the effectiveness of our proposed method.