Accurate machine learning-based germination detection, prediction and quality assessment of three grain crops

Abstract

Background Assessment of seed germination is an essential task for seed researchers to measure the quality and performance of seeds. Usually, seed assessments are done manually, which is a cumbersome, time consuming and error-prone process. Classical image analyses methods are not well suited for large-scale germination experiments, because they often rely on manual adjustments of color-based thresholds. We here propose a machine learning approach using modern artificial neural networks with region proposals for accurate seed germination detection and high-throughput seed germination experiments. Results We generated labeled imaging data of the germination process of more than 2400 seeds for three different crops, Zea mays (maize), Secale cereale (rye) and Pennisetum glaucum (pearl millet), with a total of more than 23,000 images. Different state-of-the-art convolutional neural network (CNN) architectures with region proposals have been trained using transfer learning to automatically identify seeds within petri dishes and to predict whether the seeds germinated or not. Our proposed models achieved a high mean average precision (mAP) on a hold-out test data set of approximately 97.9%, 94.2% and 94.3% for Zea mays, Secale cereale and Pennisetum glaucum respectively. Further, various single-value germination indices, such as Mean Germination Time and Germination Uncertainty, can be computed more accurately with the predictions of our proposed model compared to manual countings. Conclusion Our proposed machine learning-based method can help to speed up the assessment of seed germination experiments for different seed cultivars. It has lower error rates and a higher performance compared to conventional and manual methods, leading to more accurate germination indices and quality assessments of seeds.

Deep Learning for Image-Based Plant Growth Monitoring: A Review

“…In comparison to the pure CNN network formulated for different functions, the use of the hybrid network for plant growth monitoring is relatively new [73]. However, the incorporation of temporal elements into spatial-based CNNs has been identified as one of the promising directions for future work related to plant growth monitoring studies, highlighting the potentially broad application of hybrid networks [74][75].…”

“…), fruits at the harvest stage (e.g., date [83], blueberry [84], tomato [85], apple [86], etc. ), and seed germination (e.g., maize, rye, pearl millet [74], etc.). The works related to the classification are summarized in Table 1, which also provides an overview of the plant growth monitoring approaches reported in the reviewed studies.…”

“…This may be done via a direct comparison of the final output with the manually obtained output. For example, the germination indices obtained from the object detection task in the work of Genze et al [74] were used for cross-checking with manual measurements besides the performance in germination detection. As another example, the estimated heading date output from a model was compared against the heading date determined via observation [81].…”

Sustainable Agriculture through Multidisciplinary Seed Nanopriming: Prospects of Opportunities and Challenges

“…The quality of seeds and their vigor can be accessed through digital analysis, but also biological experiments under varying conditions can be quantitatively compared to increase the confidence of research results. A study by Genze et al presented accurate germination detection, prediction and quality assessment based on machine learning [ 96 ]. Several attempts have been made to automate seed testing in order to reduce the number of error-prone manual steps required in this process.…”

Fast anther dehiscence state recognition system establishing by deep learning to screen heat tolerant cotton

“…Therefore, the system can be used in real time in a real cultivation environment (Ferentinos, 2018) recognition system that can automatically identify seed categories (including maize, rye, and pearl millet) in petri dishes and automatically determine whether the seeds are germinating. The system achieves an average accuracy of 94% on test data and can help seed researchers to better determine seed quality and performance (Genze et al, 2020). Scientists use hyperspectral imaging technology to collect spectral and image information from maize seeds and combine convolutional neural networks and support vector machines to model and train spectral data sets and image data sets.…”