Soybean Yield Preharvest Prediction Based on Bean Pods and Leaves Image Recognition Using Deep Learning Neural Network Combined With GRNN

Soybean yield is a highly complex trait determined by multiple factors such as genotype, environment, and their interactions. The earlier the prediction during the growing season the better. Accurate soybean yield prediction is important for germplasm innovation and planting environment factor improvement. But until now, soybean yield has been determined by weight measurement manually after soybean plant harvest which is time-consuming, has high cost and low precision. This paper proposed a soybean yield in-field prediction method based on bean pods and leaves image recognition using a deep learning algorithm combined with a generalized regression neural network (GRNN). A faster region-convolutional neural network (Faster R-CNN), feature pyramid network (FPN), single shot multibox detector (SSD), and You Only Look Once (YOLOv3) were employed for bean pods recognition in which recognition precision and speed were 86.2, 89.8, 80.1, 87.4%, and 13 frames per second (FPS), 7 FPS, 24 FPS, and 39 FPS, respectively. Therefore, YOLOv3 was selected considering both recognition precision and speed. For enhancing detection performance, YOLOv3 was improved by changing IoU loss function, using the anchor frame clustering algorithm, and utilizing the partial neural network structure with which recognition precision increased to 90.3%. In order to improve soybean yield prediction precision, leaves were identified and counted, moreover, pods were further classified as single, double, treble, four, and five seeds types by improved YOLOv3 because each type seed weight varies. In addition, soybean seed number prediction models of each soybean planter were built using PLSR, BP, and GRNN with the input of different type pod numbers and leaf numbers with which prediction results were 96.24, 96.97, and 97.5%, respectively. Finally, the soybean yield of each planter was obtained by accumulating the weight of all soybean pod types and the average accuracy was up to 97.43%. The results show that it is feasible to predict the soybean yield of plants in situ with high precision by fusing the number of leaves and different type soybean pods recognized by a deep neural network combined with GRNN which can speed up germplasm innovation and planting environmental factor optimization.

Changes in soybean cultivars released over the past 50 years in southern Brazil

On-farm soybean yield has increased considerably in the last 50 years in southern Brazil, but there is still little information about how selection and breeding for yield increase has changed the agronomic attributes of cultivars. The objectives of this study were to evaluate the changes in soybean yield, seed oil and protein concentration, and changes in plant attributes that might be associated with yield improvement of 26 soybean cultivars released over the past 50 years in southern Brazil, sown simultaneously in a common field environment for two growing seasons. The average rate of yield gain was 45.9 kg ha−1 yr−1 (2.1% ha−1 yr−1), mainly due increased seed number per area and harvest index. Over year of cultivar release, cultivars became less susceptible to lodging, as well as plant mortality reduced. Meanwhile, the seed oil concentration increased, and seed protein concentration decreased, which could have negative consequences for soybeans use and requires further attention for breeding of future cultivars. Breeders have successfully contributed to the annual rate of soybean yield increase in southern Brazil. By our results, as well as the official on-farm production data, there is no evidence of soybean yield reaching a plateau in the near future in southern Brazil.

Improve Soybean Variety Selection Accuracy Using UAV-Based High-Throughput Phenotyping Technology

The efficiency of crop breeding programs is evaluated by the genetic gain of a primary trait of interest, e.g., yield, achieved in 1 year through artificial selection of advanced breeding materials. Conventional breeding programs select superior genotypes using the primary trait (yield) based on combine harvesters, which is labor-intensive and often unfeasible for single-row progeny trials (PTs) due to their large population, complex genetic behavior, and high genotype-environment interaction. The goal of this study was to investigate the performance of selecting superior soybean breeding lines using image-based secondary traits by comparing them with the selection of breeders. A total of 11,473 progeny rows (PT) were planted in 2018, of which 1,773 genotypes were selected for the preliminary yield trial (PYT) in 2019, and 238 genotypes advanced for the advanced yield trial (AYT) in 2020. Six agronomic traits were manually measured in both PYT and AYT trials. A UAV-based multispectral imaging system was used to collect aerial images at 30 m above ground every 2 weeks over the growing seasons. A group of image features was extracted to develop the secondary crop traits for selection. Results show that the soybean seed yield of the selected genotypes by breeders was significantly higher than that of the non-selected ones in both yield trials, indicating the superiority of the breeder's selection for advancing soybean yield. A least absolute shrinkage and selection operator model was used to select soybean lines with image features and identified 71 and 76% of the selection of breeders for the PT and PYT. The model-based selections had a significantly higher average yield than the selection of a breeder. The soybean yield selected by the model in PT and PYT was 4 and 5% higher than those selected by breeders, which indicates that the UAV-based high-throughput phenotyping system is promising in selecting high-yield soybean genotypes.

Impacts of row spacing and fungicide timing on foliar disease, greenstem and yield in double cropped soybeans grown in the Chesapeake Bay region of the United States.

Double cropped soybeans are planted on approximately 1/3 of crop acres in the Chesapeake Bay region of the United States. Producers have asked if foliar fungicides are required to optimize yields in this region. We assessed the impacts of foliar fungicide application timing and row spacing on foliar disease, greenstem, and yield from 11 site years spanning 2017-2019. Foliar diseases only developed at rateable levels in one location. Fungicide application, regardless of timing, increased percent greenstem over non-treated controls. Fungicide application did not impact soybean yield. Yield was greater in 38.1 cm rows when compared to 19 cm rows. Our data do not support the use of foliar fungicides in double cropped soybean production in this region.