Longgang Zhao

Phenotypic characterization of mature soybean pods is a crucial aspect of breeding programs, yet efficiently obtaining accurate pod phenotypic parameters remains a major challenge. Recent advances in deep learning, particularly in keypoint detection models, have introduced innovative methods for pod phenotype extraction. However, precise identification and analysis of fine-scale phenotypic traits in soybean pods remain challenging in current research. We propose Pod-pose, an innovative top-down keypoint detection model for precise soybean pod phenotyping that adapts human pose estimation techniques to plant phenotyping. Specifically, Pod-pose integrates the architectural strengths of various advanced YOLO (You Only Look Once) models through bottleneck structure optimization and positional feature enhancement to achieve superior detection accuracy. Furthermore, we implemented a two-stage detection method augmented with transfer learning, which not only reduces training complexity but also significantly enhances the model's performance. Extensive evaluation of our custom-built dataset demonstrated Pod-Pose's superior performance, with the X variant achieving an Average Precision of 0.912 at an IoU threshold of 0.5 (AP@IoU = 0.5). Notably, four critical pod-related phenotypic traits were successfully quantified: pod length, bending length, curvature, and inflection point width. This study establishes Pod-Pose as a viable solution for pod phenotyping, with potential applications in soybean breeding optimization.

The soybean stem node is a key part of soybean growth and development, and its numbers play a crucial role in soybean yield formation. Traditional manual methods are labor-intensive and error-prone. The keypoint detection method is an ideal choice for stem node detection due to its high accuracy and wide applicability. In this study, a new deep learning method, You Only Look Once _Soybean Stalk Pose (YOLO_SSP) was proposed, which innovatively applied the Small_Effective Low-Level Aggregation Network (S_ELAN) module and fused it with a smaller detection head for detecting stem nodes in mature soybeans. After optimization and iteration, the model achieved 88.1% accuracy on the dataset. Subsequently, by ablating the model, it was found that different improvements were effective in increasing the accuracy of the model. In addition, when comparing the classic YOLO series of keypoint detection models, the results show that YOLO_SSP achieved up to 87.7% of APs, which was higher than YOLOv7-w6-pose, YOLOv7-tiny-pose, YOLOv3s-pose, YOLOv5n-pose, YOLOv5s-pose, YOLOv5m- pose, YOLOv6n-pose, YOLOv8n-pose, and YOLOv10b-pose, which were 2.5%, 12.8%, 5.3%, 3.8%, 3.5%, 3.5%, 5.1%, 5.1%, 5.0%, and 4.5% higher, respectively. Finally, the proposed model was applied to the unique dataset with 85.3% precision and 82.6% accuracy, and the visualization of the model’s detection results proved its applicability and universality. This study provides an effective strategy for soybean stem node detection and significantly improves the accuracy of detection.

Accurate soybean pod counting remains a significant challenge in field-based phenotyping due to complex factors such as occlusion, dense distributions, and background interference. We present SmartPod, an advanced deep learning framework that addresses these challenges through three key innovations: (1) a novel vision Transformer architecture for enhanced feature representation, (2) an efficient attention mechanism for the improved detection of overlapping pods, and (3) a semi-supervised learning strategy that maximizes performance with limited annotated data. Extensive evaluations demonstrate that SmartPod achieves state-of-the-art performance with an Average Precision at an IoU threshold of 0.5 (AP@IoU = 0.5) of 94.1%, outperforming existing methods by 1.7–4.6% across various field conditions. This significant improvement, combined with the framework’s robustness in complex environments, positions SmartPod as a transformative tool for large-scale soybean phenotyping and precision breeding applications.

Salt stress impedes normal development, compromises plant quality, and reduces crop yield. The germination phase in soybean marks the initial stage of its growth cycle. Characterizing salt tolerance during this period can help stimulate soybean growth in natural environments and aid the rapid screening of salt-tolerant soybean varieties. Our study characterized the salt tolerance of 36 soybean germplasms in culture dishes during the germination period. Soybeans were subjected to varying concentrations (0, 60, 120, and 180 mmol/L) of NaCl solution to simulate diverse levels of salt stress, and parameters such as germination energy, germination rate, and root length were measured. Statistical techniques such as analysis of variance, membership function, cluster analysis, and quadratic regression equations were used, and the salt tolerance of these 36 soybean germplasms was determined. The critical indicators and the most effective screening concentration for assessing the germination salt tolerance of soybean were identified. Soybeans tolerated low salt concentrations; however, salt concentrations greater than 120 mmol/L significantly inhibited germination indicators. The germination rate, germination vigor, vitality index, seed germination index, total fresh weight, and total dry weight could be used to identify salt tolerance. The semi-lethal concentration of soybean was 155.4 mmol/L, and the coefficient of variation was 20.00%, indicating that it could be used as a screening concentration for evaluating salt tolerance during soybean germination. A total of 36 soybean varieties were classified into four salt tolerance levels through cluster analysis. QN-27, QN-35, and QN-36 were highly salt-resistant materials, and QN-2, QN-17, and QN-19 were salt-sensitive materials. Characterizing salt tolerance during soybean germination can facilitate the selection and breeding of salt-tolerant soybean varieties. Future research utilizing this approach can aid in the selection of soybean varieties with salinity tolerance.

The cultivation of saline–alkali-tolerant peanut (Arachis hypogaea L.) varieties can effectively increase grain yield in saline–alkali land. However, traditional assessment methods are often cumbersome and time consuming. To rapidly identify saline–alkali stress-tolerant peanut varieties, this research proposed a saline–alkali stress tolerance evaluation method based on deep learning and multimodal data. Specifically, the research first established multimodal datasets for peanuts at different growth stages and constructed a saline–alkali stress score standard based on unsupervised learning. Subsequently, a deep learning network called BO-MFFNet was built and its structure and hyperparameters were optimized by the Bayes optimization (BO) algorithm. Finally, the point prediction of the saline–alkali stress score were carried out by using the Gaussian process regression model. The experimental results show that the multimodal method is superior to the single-modal data and the BO algorithm significantly improves the performance of the model. The root mean squared error and relative percentage deviation of the BO-MFFNet model are 0.089 and 3.669, respectively. The model effectively predicted the salt–alkali stress tolerance of five varieties, and the predicted results were Huayu25, Yuhua31, Yuhua33, Yuhua32, and Yuhua164 from high to low. This research provides a new method for assessing crop tolerance under extreme environmental stress.

Various biotic and abiotic stresses challenge plant growth and production, with salt stress notably impeding normal development, compromising quality, and reducing yield in plants. The objective is to identify salt-tolerant soybean germplasm resources and develop a comprehensive method for assessing salt tolerance during soybean germination. This study involved the utilization of 36 soybean germplasm to induce varying degrees of salt stress through the application of 0, 60, 120, and 180 mmol/L NaCl solutions, employing the Petri dish germination identification method to assess nine phenotypic indices, such as germination potential, rate, and radicle length. The salt tolerance of soybean germination was comprehensively assessed using one-way ANOVA, multi-factor ANOVA, principal component analysis, affiliation function, quadratic regression equation analysis, and cluster analysis. These methods were employed to investigate the impact of varied NaCl concentrations on different germination indices, determine the best screening parameters for identifying soybean germination salt tolerance, and identify soybean germplasm with varying salt tolerances. The results indicated that soybean seeds exhibited tolerance to low salt concentrations, while concentrations exceeding 120 mmol/L led to significant inhibition of germination indicators. Notably, the vitality index and radicle length were highly sensitive indicators reflecting the impact of salt stress on seed germination. By analyzing the quadratic regression equation correlating the germination index and the salt tolerance coefficient, an LC50 value of 155.4 mmol/L with a coefficient of variation of 20.00% was derived. This value serves as a viable screening concentration for identifying salt tolerance during the soybean germination stage. Additionally, clustering analysis categorized 36 soybean varieties into 4 salt tolerance levels, with QN-27, QN-35, and QN-36 demonstrating high salt tolerance, and QN-2, QN-17, and QN-19 were classified as salt-sensitive materials.