High-throughput soybean seeds phenotyping with convolutional neural networks and transfer learning

Abstract

Background Effective soybean seed phenotyping demands large-scale accurate quantities of morphological parameters. The traditional manual acquisition of soybean seed morphological phenotype information is error-prone, and time-consuming, which is not feasible for large-scale collection. The segmentation of individual soybean seed is the prerequisite step for obtaining phenotypic traits such as seed length and seed width. Nevertheless, traditional image-based methods for obtaining high-throughput soybean seed phenotype are not robust and practical. Although deep learning-based algorithms can achieve accurate training and strong generalization capabilities, it requires a large amount of ground truth data which is often the limitation step. Results We showed a novel synthetic image generation and augmentation method based on domain randomization. We synthesized a plenty of labeled image dataset automatedly by our method to train instance segmentation network for high throughput soybean seeds segmentation. It can pronouncedly decrease the cost of manual annotation and facilitate the preparation of training dataset. And the convolutional neural network can be purely trained by our synthetic image dataset to achieve a good performance. In the process of training Mask R-CNN, we proposed a transfer learning method which can reduce the computing costs significantly by finetuning the pre-trained model weights. We demonstrated the robustness and generalization ability of our method by analyzing the result of synthetic test datasets with different resolution and the real-world soybean seeds test dataset. Conclusion The experimental results show that the proposed method realized the effective segmentation of individual soybean seed and the efficient calculation of the morphological parameters of each seed and it is practical to use this approach for high-throughput objects instance segmentation and high-throughput seeds phenotyping.

A Deep-Learning-Based Approach for Wheat Yellow Rust Disease Recognition from Unmanned Aerial Vehicle Images

“…Liang et al [ 44 ] proposed a recognition model with a multilevel convolutional feature pyramid to solve the problem of poor generalization in the recognition and classification of food images. Yang et al [ 45 ] used convolutional neural networks for training, and demonstrated the stability and generalization of the method by analyzing the training results of a synthetic dataset and a dataset of real soybean seeds. In this study, the semantic segmentation model PSPNet was trained using the UAV images and corresponding category labels of Xigolou Village in Yuzhou City, Henan Province in 2021 as the training set.…”

High-Throughput Legume Seed Phenotyping Using a Handheld 3D Laser Scanner

“…Legume seeds are rigid. The shape of legume seeds, such as soybeans, peas, black beans, red beans, and mung beans, are approximately spherical or ellipsoidal [34,35,41], meaning they are almost symmetrical. Therefore, this paper exploits the geometric symmetry characteristics of legume seeds to reconstruct the 3D model based on the scanned incomplete point cloud.…”

“…Examples of spherical legume seeds are soybeans and peas [34]. Examples of ellipsoidal legume seeds are black beans, red beans, and mung beans [35]. Spherical or elliptical seeds are symmetrical, a property that can be taken advantage of for rapid batch 3D modeling.…”

A novel image-based approach for soybean seed phenotyping using machine learning techniques

“…In the last few decades, image analysis software programs for seed evaluation have emerged with the use of well-established image processing methodologies, allowing the evaluation of several morphological characteristics, which have been little explored due to the difficulty of manual measurements, such as perimeter and circularity (Lamprecht et al, 2007;Igathinathane et al, 2008;Tanabata et al, 2012;Gehan et al, 2017). However, such approaches have shown diverse limitations, especially for segmenting seeds under inconsistent lighting conditions or that are densely arranged and have physical contact with each other (Yang et al, 2021). When seeds are closely placed, they are often detected as a unified image component, implying the misrecognition of isolated seeds (Toda et al, 2020).…”

“…Even with such importance, little efforts to solve this problem have been reported in the literature. Although Yang et al (2021) proposed a deep learning-based method for image segmentation in seeds arranged with physical contact, the authors report a high computational cost and the need of a large amount of data to train the model, hindering the application in smaller datasets. Our image segmentation framework proposed here, on the other hand, is entirely based on mathematical morphological operations, which resulted in an easy implementation pipeline with low computational costs.…”