High-throughput image-based plant stand count estimation using convolutional neural networks

Khaki, Saeed; Pham, Hieu; Khalilzadeh, Zahra; Masoud, Ahmad A.; Safaei, Nima; Han, Ye; Kent, Wade; Wang, Lizhi

doi:10.1371/journal.pone.0268762

Cited by 5 publications

(4 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If the images are acquired too early, the seedlings are too small to be detected and counted, e.g., seedling detection and counting at T1 in this study. If the images are acquired too late, the cotton seedlings reach the four-leaf stage, and the leaves have spread out, with crossing and overlap occurring between leaves, resulting in underestimation of the cotton population, e.g., seedling detection and counting at T6 in this study, which are also the two significant challenges encountered so far [51]. Due to the short window of cotton seedlings, if the best time to replenish them is missed, it will affect the cotton yield at the later harvest.…”

Section: Discussionmentioning

confidence: 98%

Cotton Seedling Detection and Counting Based on UAV Multispectral Images and Deep Learning Methods

Feng

Chen

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Cotton is one of the most important cash crops in Xinjiang, and timely seedling inspection and replenishment at the seedling stage are essential for cotton’s late production management and yield formation. The background conditions of the cotton seedling stage are complex and variable, and deep learning methods are widely used to extract target objects from the complex background. Therefore, this study takes seedling cotton as the research object and uses three deep learning algorithms, YOLOv5, YOLOv7, and CenterNet, for cotton seedling detection and counting using images at six different times of the cotton seedling period based on multispectral images collected by UAVs to develop a model applicable to the whole cotton seedling period. The results showed that when tested with data collected at different times, YOLOv7 performed better overall in detection and counting, and the T4 dataset performed better in each test set. Precision, Recall, and F1-Score values with the best test results were 96.9%, 96.6%, and 96.7%, respectively, and the R2, RMSE, and RRMSE indexes were 0.94, 3.83, and 2.72%, respectively. In conclusion, the UAV multispectral images acquired about 23 days after cotton sowing (T4) with the YOLOv7 algorithm achieved rapid and accurate seedling detection and counting throughout the cotton seedling stage.

show abstract

Section: Discussionmentioning

confidence: 98%

Cotton Seedling Detection and Counting Based on UAV Multispectral Images and Deep Learning Methods

Feng

Chen

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…From the literature, it was found that there are many ML studies [10][11][12][13][14] and DL studies [39-43] have been proposed for finding and counting oil palm trees, to estimate yields, monitor crops, and identify nutritional deficiencies etc. However, comprehensive (monitoring growth and counting the healthy and unhealthy plantlings of palm tree with high accuracy) study on palm tree plantlings is yet to be researched.…”

Section: Plos Onementioning

confidence: 99%

“…Numerous research on useful resources and ML methods for the palm oil sector was undertaken. Research has investigated the use of remote sensing, breeding, and technology to keep tabs on palm oil farms [11]. Bioenergy manufacturing technologies for dealing with fruits and palm oil waste have been evaluated in another research [12].…”

Section: Introductionmentioning

confidence: 99%

A novel CNN gap layer for growth prediction of palm tree plantlings

Kumar,

Rajmohan,

Adeola Ajagbe

et al. 2023

PLoS ONE

View full text Add to dashboard Cite

Monitoring palm tree seedlings and plantlings presents a formidable challenge because of the microscopic size of these organisms and the absence of distinguishing morphological characteristics. There is a demand for technical approaches that can provide restoration specialists with palm tree seedling monitoring systems that are high-resolution, quick, and environmentally friendly. It is possible that counting plantlings and identifying them down to the genus level will be an extremely time-consuming and challenging task. It has been demonstrated that convolutional neural networks, or CNNs, are effective in many aspects of image recognition; however, the performance of CNNs differs depending on the application. The performance of the existing CNN-based models for monitoring and predicting plantlings growth could be further improved. To achieve this, a novel Gap Layer modified CNN architecture (GL-CNN) has been proposed with an IoT effective monitoring system and UAV technology. The UAV is employed for capturing plantlings images and the IoT model is utilized for obtaining the ground truth information of the plantlings health. The proposed model is trained to predict the successful and poor seedling growth for a given set of palm tree plantling images. The proposed GL-CNN architecture is novel in terms of defined convolution layers and the gap layer designed for output classification. There are two 64×3 conv layers, two 128×3 conv layers, two 256×3 conv layers and one 512×3 conv layer for processing of input image. The output obtained from the gap layer is modulated using the ReLU classifier for determining the seedling classification. To evaluate the proposed system, a new dataset of palm tree plantlings was collected in real time using UAV technology. This dataset consists of images of palm tree plantlings. The evaluation results showed that the proposed GL-CNN model performed better than the existing CNN architectures with an average accuracy of 95.96%.

show abstract

“…Examples of this cross-over are presented in Table 2 , where despite semantic differences, the research fundamentals between plant science and weed recognition are common. Khaki et al (2022a) recognise this opportunity for the generalizability of a stand count approach in maize to contribute to weed detection, which could be combined with works such as Picon et al (2022) for maize extraction within weedy environments. Recognising the potential for research overlap, Weyler et al (2021) present a crop-weed detection system, which incorporates leaf counting for growth stage estimation and in-field phenotyping.…”

Section: Similarities Between Phenotyping and Weed Recognitionmentioning

confidence: 99%

More eyes on the prize: open-source data, software and hardware for advancing plant science through collaboration

Coleman

Salter

2023

AoB PLANTS

View full text Add to dashboard Cite

Automating the analysis of plants using image processing would help remove barriers to phenotyping and large-scale precision agricultural technologies, such as site-specific weed control. The combination of accessible hardware and high-performance deep learning (DL) tools for plant analysis is becoming widely recognised as a path forward for both plant science and applied precision agricultural purposes. Yet, a lack of collaboration in image analysis for plant science, despite the open-source origins of much of the technology, is hindering development. Here, we show how tools developed for specific attributes of phenotyping or weed recognition for precision weed control have substantial overlapping data structure, software/hardware requirements and outputs. An open-source approach to these tools facilitates interdisciplinary collaboration, avoiding unnecessary repetition and allowing research groups in both basic and applied sciences to capitalise on advancements and resolve respective bottlenecks. The approach mimics that of machine learning in its nascence. Three areas of collaboration are identified as critical for improving efficiency, (1) standardised, open-source, annotated dataset development with consistent metadata reporting; (2) establishment of accessible and reliable training and testing platforms for DL algorithms; and (3) sharing of all source code used in the research process. The complexity of imaging plants and cost of annotating image datasets means that collaboration from typically distinct fields will be necessary to capitalise on the benefits of DL for both applied and basic science purposes.

show abstract

High-throughput image-based plant stand count estimation using convolutional neural networks

Cited by 5 publications

References 52 publications

Cotton Seedling Detection and Counting Based on UAV Multispectral Images and Deep Learning Methods

Cotton Seedling Detection and Counting Based on UAV Multispectral Images and Deep Learning Methods

A novel CNN gap layer for growth prediction of palm tree plantlings

More eyes on the prize: open-source data, software and hardware for advancing plant science through collaboration

Contact Info

Product

Resources

About