Sinohyriopsis cumingii is an economically important freshwater mussel endemic to China and plays a core role in the

S. cumingii is an economically important freshwater mussel endemic to China and plays a core role in the global pearl aquaculture industry. Phenotypic traits of S. cumingii are fundamental and essential for evaluating individual growth performance, identifying germplasm resources, and implementing precise genetic breeding programs. However, conventional manual measurement methods are labor-intensive, time-consuming, and highly susceptible to subjective human errors, often requiring skilled technicians and introducing inconsistent measurements between different operators, which severely limit their scalability and applicability in large-scale intelligent aquaculture production. To address these critical limitations and achieve non-destructive, rapid, and accurate acquisition of phenotypic parameters, this study proposes an improved measurement method based on YOLOv8n, termed YOLOv8n-CBM.First, an integrated automatic phenotypic measurement system for S. cumingii was constructed by combining dynamic transmission, machine vision, and digital image processing technologies. The system comprises a conveyor device, a high-precision industrial camera, and a dedicated image processing module, which automatically transports samples to the imaging area, enabling standardized image acquisition and high-throughput phenotypic measurement of mussel samples under continuous dynamic operating conditions. Then, three targeted improvements were implemented in the original YOLOv8n network architecture to adapt to the characteristics of mussel images. In the backbone network, four convolutional block attention modules (CBAM) were embedded after each C2f block to enhance the extraction of contour edges and local detailed features of mussels while effectively suppressing irrelevant background interference. In the neck network, the bidirectional feature pyramid network (BiFPN) was introduced to strengthen bidirectional fusion of multi-scale features and improve the model’s representation ability for targets of different sizes and postures. Meanwhile, the original C2f module was replaced with a multi-scale dilated attention (MSDA) module to expand the network’s receptive field and enhance the joint modeling capability of local fine-grained details and global contextual information. Finally, key phenotypic parameters including shell length, full height, shell height, and radial rib length of the buttock angle were extracted using the geometric relationship between rotated bounding boxes and biological key points. Experiments were conducted on a dataset of 50 manually annotated S. cumingii samples covering diverse sizes and postures. The results show that the mean average precision (mAP50-95) of the YOLOv8n-CBM model reached 98.2%, demonstrating a notable improvement in rotated object detection performance over the original YOLOv8n model. The average localization deviation of biological key points was less than 2.0 mm, indicating high precision in anatomical feature detection. The mean absolute errors (MAE) of shell length, full height, shell height, and radial rib length of the buttock angle were 1.51, 1.08, 1.02, and 2.00 mm, respectively, with all relative errors remaining below 2%. In all groups stratified by different shell lengths and full heights, the measurement errors of YOLOv8n-CBM were consistently lower than those of the original YOLOv8n model, with the maximum absolute error strictly controlled within 2.879 mm. The proposed model effectively improved measurement accuracy and robustness for S. cumingii with diverse morphologies and postures. In conclusion, this method effectively overcomes the inherent drawbacks of traditional manual measurement and realizes rapid, accurate, and non-destructive acquisition of phenotypic traits in S. cumingii. It provides a reliable technical approach for shellfish growth evaluation and genetic breeding, supports the transition of the pearl industry from empirical farming to data-driven and intelligent aquaculture, and offers a valuable reference for automated phenotypic measurement in other molluscan species.