Shijing Liu

Phenotyping technologies are essential for modern aquaculture, particularly for precise analysis of individual morphological traits. This study focuses on critical phenotype segmentation tasks for fish carcass and fins, which have significant applications in phenotypic assessment and breeding. In high-density underwater environments, fish frequently exhibit structural overlap and indistinct boundaries, making it difficult for conventional segmentation methods to obtain complete and accurate phenotypic regions. To address these challenges, a double-branch segmentation network is proposed for fish phenotype segmentation in high-density underwater scenes. An auxiliary saliency object detection (SOD) branch is introduced alongside the primary segmentation branch to localize structurally complete targets and suppress interference from overlapping or incomplete fish while inter-branch skip connections further enhance the model’s focus on salient targets and their boundaries. The network is trained under a multi-task learning framework, allowing the branches to specialize in edge detection and accurate region segmentation. Experiments on large yellow croaker (Larimichthys crocea) images collected under real farming conditions show that the proposed method achieves Dice scores of 97.58% for carcass segmentation and 88.88% for fin segmentation. The corresponding ASD values are 0.590 and 0.364 pixels, and the HD95 values are 3.521 and 1.222 pixels. The method outperforms nine existing algorithms across key metrics, confirming its effectiveness and reliability for practical aquaculture phenotyping.

Precise identification and quantification of fish movement states are of significant importance for conducting fish behavior research and guiding aquaculture production, with object tracking serving as a key technical approach for achieving behavioral quantification. The traditional DeepSORT algorithm has been widely applied to object tracking tasks; however, in practical aquaculture environments, high-density cultured fish exhibit visual characteristics such as similar textural features and frequent occlusions, leading to high misidentification rates and frequent ID switching during the tracking process. This study proposes an underwater fish object tracking method based on the improved DeepSORT algorithm, utilizing ResNet as the backbone network, embedding Deformable Convolutional Networks v2 to enhance adaptive receptive field capabilities, introducing Triplet Loss function to improve discrimination ability among similar fish, and integrating Convolutional Block Attention Module to enhance key feature learning. Finally, by combining the aforementioned improvement modules, the ReID feature extraction network was redesigned and optimized. Experimental results demonstrate that the improved algorithm significantly enhances tracking performance under frequent occlusion conditions, with the MOTA metric improving from 64.26% to 66.93% and the IDF1 metric improving from 53.73% to 63.70% compared to the baseline algorithm, providing more reliable technical support for underwater fish behavior analysis.

Introduction: The eating sounds of largemouth black bass (Micropterus salmoides) are primarily categorized into swallowing and chewing sounds, both intensities of which are closely correlated with fish density and feeding desire. Therefore, accurate recognition of these two sounds is of significant importance for studying fish feeding behavior.Methods: In this study, we propose a method based on low-dimensional acoustic features for the recognition of swallowing and chewing sounds in fish. Initially, utilizing synchronous audio-visual means, we collect feeding sound signals and image signals of largemouth black bass. By analyzing the time-frequency domain features of the sound signals, we identify 15 key acoustic features across four categories including short-time average energy, average Mel-frequency cepstral coefficients, power spectral peak, and center frequency. Subsequently, employing nine dimensionality reduction algorithms, we select the Top-6 features from the 15-dimensional acoustic features and compare their precision in recognizing swallowing and chewing sounds using four machine learning models.Results: Experimental results indicate that supervised feature pre-screening positively enhances the accuracy of largemouth black bass feeding feature recognition. Extracted acoustic features demonstrate global correlation and linear characteristics. When considering feature dimensionality and classification performance, the combination of feature dimensionality reduction and recognition model based on the random forest model exhibits the best performance, achieving an identification accuracy of 98.63%.Discussion: The proposed method offers higher assessment accuracy of swallowing and chewing sounds with lower computational complexity, thus providing effective technical support for the research on precise feeding technology in fish farming.

Variable-condition fish recognition is a type of cross-scene and cross-camera fish re-identification (re-ID) technology. Due to the difference in the domain distribution of fish images collected under different culture conditions, the available training data cannot be effectively used for the new identification method. To solve these problems, we proposed a new method for identifying large yellow croaker based on the CycleGAN (cycle generative adversarial network) and transfer learning. This method constructs source sample sets and target sample sets by acquiring large yellow croaker images in controllable scenes and actual farming conditions, respectively. The CycleGAN was used as the basic framework for image transformation from the source domain to the target domain to realize data amplification in the target domain. In particular, IDF (identity foreground loss) was used to optimize identity loss judgment criteria, and MMD (maximum mean discrepancy) was used to narrow the distribution between the source domain and target domain. Finally, transfer learning was carried out with the expanded samples to realize the identification of large yellow croaker under varying conditions. The experimental results showed that the proposed method achieved good identification results in both the controlled scene and the actual culture scene, with an average recognition accuracy of 96.9% and 94%, respectively. These provide effective technical support for the next steps in fish behavior tracking and phenotype measurement.

The intensity and frequency of the acoustic signals generated by different behaviors of largemouth bass (Micropterus salmoides) have different characteristics. The acoustic signals generated during feeding can be used to analyze the characteristic patterns of their used behavior, which can provide a theoretical basis for applications such as automatic feeding based on acoustic signals. We used passive acoustics combined with video to study the feeding acoustic signals of largemouth bass in a recirculating water culture system (4, 8, 12, and 16 fish/m³). The result of the time–frequency and power spectrum analysis of the feeding sound signals showed that the short-time average amplitude of the feeding signal was well distinguished from the background noise, both the swallowing and the chewing sounds were positively correlated with the culture density, and the correlation between the swallowing sound and number of fish was stronger; at different culture densities, the short-time average zero-crossing in the swallowing phase of the largemouth bass suddenly dropped to about 500 and rose to about 1000 in the chewing process. Therefore, both the chewing and the swallowing sounds can be used as parameters to automatically identify the largemouth bass feeding process signal. The spectrum entropy in the feeding process of the largemouth bass was maintained at about 4–6 for different culture densities. In the power spectrum, the main feeding sounding frequencies of the largemouth bass at different farming densities and the distinguishable spectral range of the background noised ranged from 1 to 20 kHz, the main peak frequency of the feeding sound was within the range of 1.2 to 3.0 kHz, and the main power value of the power spectrum was positively correlated with farming density.