The number of panicles per unit area is a common indicator of rice yield and is of great significance to yield estimation, breeding, and phenotype analysis. Traditional counting methods have various drawbacks, such as long delay times and high subjectivity, and they are easily perturbed by noise. To improve the accuracy of rice detection and counting in the field, we developed and implemented a panicle detection and counting system that is based on improved region-based fully convolutional networks, and we use the system to automate rice-phenotype measurements. The field experiments were conducted in target areas to train and test the system and used a rotor light unmanned aerial vehicle equipped with a high-definition RGB camera to collect images. The trained model achieved a precision of 0.868 on a held-out test set, which demonstrates the feasibility of this approach. The algorithm can deal with the irregular edge of the rice panicle, the significantly different appearance between the different varieties and growing periods, the interference due to color overlapping between panicle and leaves, and the variations in illumination intensity and shading effects in the field. The result is more accurate and efficient recognition of rice-panicles, which facilitates rice breeding. Overall, the approach of training deep learning models on increasingly large and publicly available image datasets presents a clear path toward smartphone-assisted crop disease diagnosis on a global scale.

Leaf coverage is an indicator of plant growth rate and predicted yield, and thus it is crucial to plant-breeding research. Robust image segmentation of leaf coverage from remote-sensing images acquired by unmanned aerial vehicles (UAVs) in varying environments can be directly used for large-scale coverage estimation, and is a key component of high-throughput field phenotyping. We thus propose an image-segmentation method based on machine learning to extract relatively accurate coverage information from the orthophoto generated after preprocessing. The image analysis pipeline, including dataset augmenting, removing background, classifier training and noise reduction, generates a set of binary masks to obtain leaf coverage from the image. We compare the proposed method with three conventional methods (Hue-Saturation-Value, edge-detection-based algorithm, random forest) and a frontier deep-learning method called DeepLabv3+. The proposed method improves indicators such as Q_seg, S_r, E_s and mIOU by 15% to 30%. The experimental results show that this approach is less limited by radiation conditions, and that the protocol can easily be implemented for extensive sampling at low cost. As a result, with the proposed method, we recommend using red-green-blue (RGB)-based technology in addition to conventional equipment for acquiring the leaf coverage of agricultural crops.

In the development of agricultural picking equipment, the identification and positioning of the target fruit is one of the key technologies. There are three sequence steps: finding the target fruit, determining whether the target fruit is mature and needs to be picked, and determining the XYZ coordinate point of the target fruit. In this study, we modified a series of convolution neural networks to detect strawberries on XY scale and judge their maturity. In order to improve the precision of small object detection, we innovatively proposed a new structure called small feature extraction layers. On a moderate GPU, small feature extraction layers improved the positioning precision from 80.53% to 89.96% and small objects precision from 41.58% to 86.14% at a 110 fps object detection speed. When using binocular ranging to measure Z-axis distance, we achieved 15 fps XYZ coordinate determination speed on a 4.5W ultra low-power CPU, and the maximum Z-axis ranging error within 60 cm was less than 2.83%.

Prostrate growth 1 (PROG1) gene is vital in controlling the prostrate growth habit of rice. Studying the effect of PROG1 gene on rice canopy structure is crucial in elucidating the molecular mechanism of rice plant type evolution. Herein, the morphological characteristics of different rice genotypes were collected at different growth stages and leaf nodes using image processing techniques. The morphological characteristics included leaf length, leaf width, and leaf area. The image processing techniques involved boundary mean oscillation (BMO) filtering and minimum bounding rectangle extraction of the target image. On this basis, the effect of the PROG1 gene on rice leaf morphology was quantitatively assessed. Also, the feasibility of image processing techniques in detecting the morphological characteristics of rice leaves was discussed. Under the influence of the PROG1 gene, the length, width, and area of rice leaves decreased by 45.1%, 12.7%, and 44.8%, respectively, at the booting stage. Similarly, the length, width, and area of flag leaves decreased by 15.8%, 32.0%, and 33.7% at the heading stage and by 25.4%, 16.2%, and 19.7% at the filling stage, respectively, and that of secondary leaf reduced by 23.2%, 13.6%, and 54.2% at heading stage and by 24.1%, 17.3%, and 37.0% at filling stage, respectively. Furthermore, the length, width, and area of other leaves reduced by 32.3%, 9.8%, and 51.6% at the heading stage and by 28.6%, 7.3%, and 36.7% at the filling stage, respectively. The leaves in the rice canopy were shorter, narrower, and smaller in leaf area. Notably, no significant differences were found between image processing technology and manual measurement methods regarding the values of leaf morphological characteristics obtained ( ). Thus, these results show that image processing technology is effective in studying the morphological characteristics of rice leaves. This study provides a reliable foundation for molecular breeding studies and will guide the application of the PROG1 gene in molecular breeding.