Javascript must be enabled to continue!
Estimates of Maize Plant Density from UAV RGB Images Using Faster-RCNN Detection Model: Impact of the Spatial Resolution
View through CrossRef
Early-stage plant density is an essential trait that determines the fate of a genotype under given environmental conditions and management practices. The use of RGB images taken from UAVs may replace the traditional visual counting in fields with improved throughput, accuracy, and access to plant localization. However, high-resolution images are required to detect the small plants present at the early stages. This study explores the impact of image ground sampling distance (GSD) on the performances of maize plant detection at three-to-five leaves stage using Faster-RCNN object detection algorithm. Data collected at high resolution (GSD≈0.3 cm) over six contrasted sites were used for model training. Two additional sites with images acquired both at high and low (GSD≈0.6 cm) resolutions were used to evaluate the model performances. Results show that Faster-RCNN achieved very good plant detection and counting (rRMSE=0.08) performances when native high-resolution images are used both for training and validation. Similarly, good performances were observed (rRMSE=0.11) when the model is trained over synthetic low-resolution images obtained by downsampling the native training high-resolution images and applied to the synthetic low-resolution validation images. Conversely, poor performances are obtained when the model is trained on a given spatial resolution and applied to another spatial resolution. Training on a mix of high- and low-resolution images allows to get very good performances on the native high-resolution (rRMSE=0.06) and synthetic low-resolution (rRMSE=0.10) images. However, very low performances are still observed over the native low-resolution images (rRMSE=0.48), mainly due to the poor quality of the native low-resolution images. Finally, an advanced super resolution method based on GAN (generative adversarial network) that introduces additional textural information derived from the native high-resolution images was applied to the native low-resolution validation images. Results show some significant improvement (rRMSE=0.22) compared to bicubic upsampling approach, while still far below the performances achieved over the native high-resolution images.
American Association for the Advancement of Science (AAAS)
Title: Estimates of Maize Plant Density from UAV RGB Images Using Faster-RCNN Detection Model: Impact of the Spatial Resolution
Description:
Early-stage plant density is an essential trait that determines the fate of a genotype under given environmental conditions and management practices.
The use of RGB images taken from UAVs may replace the traditional visual counting in fields with improved throughput, accuracy, and access to plant localization.
However, high-resolution images are required to detect the small plants present at the early stages.
This study explores the impact of image ground sampling distance (GSD) on the performances of maize plant detection at three-to-five leaves stage using Faster-RCNN object detection algorithm.
Data collected at high resolution (GSD≈0.
3 cm) over six contrasted sites were used for model training.
Two additional sites with images acquired both at high and low (GSD≈0.
6 cm) resolutions were used to evaluate the model performances.
Results show that Faster-RCNN achieved very good plant detection and counting (rRMSE=0.
08) performances when native high-resolution images are used both for training and validation.
Similarly, good performances were observed (rRMSE=0.
11) when the model is trained over synthetic low-resolution images obtained by downsampling the native training high-resolution images and applied to the synthetic low-resolution validation images.
Conversely, poor performances are obtained when the model is trained on a given spatial resolution and applied to another spatial resolution.
Training on a mix of high- and low-resolution images allows to get very good performances on the native high-resolution (rRMSE=0.
06) and synthetic low-resolution (rRMSE=0.
10) images.
However, very low performances are still observed over the native low-resolution images (rRMSE=0.
48), mainly due to the poor quality of the native low-resolution images.
Finally, an advanced super resolution method based on GAN (generative adversarial network) that introduces additional textural information derived from the native high-resolution images was applied to the native low-resolution validation images.
Results show some significant improvement (rRMSE=0.
22) compared to bicubic upsampling approach, while still far below the performances achieved over the native high-resolution images.
Related Results
Intelligent image analysis for retrieval of leaf chlorophyll content of rice from digital images of smartphone under natural light
Intelligent image analysis for retrieval of leaf chlorophyll content of rice from digital images of smartphone under natural light
The present study describes a new imaging method to acquire rice leaf images under field conditions using a smartphone and modeling approaches to retrieve the leaf chlorophyll (Chl...
A Statistical Growth Property of Plant Root Architectures
A Statistical Growth Property of Plant Root Architectures
Numerous types of biological branching networks, with varying shapes and sizes, are used to acquire and distribute resources. Here, we show that plant root and shoot architectures ...
KAT4IA: K-Means Assisted Training for Image Analysis of Field-Grown Plant Phenotypes
KAT4IA: K-Means Assisted Training for Image Analysis of Field-Grown Plant Phenotypes
High-throughput phenotyping enables the efficient collection of plant trait data at scale. One example involves using imaging systems over key phases of a crop growing season. Alth...
The characteristics and influencing factors of the spatial distribution of intangible cultural heritage in the Yellow River Basin of China
The characteristics and influencing factors of the spatial distribution of intangible cultural heritage in the Yellow River Basin of China
AbstractThis paper explores the characteristics and influencing factors of the spatial distribution of 889 national intangible cultural heritage sites in the Yellow River basin of ...
Population Flow and Epidemic Spread: Direct Impact and Spatial Spillover Effect
Population Flow and Epidemic Spread: Direct Impact and Spatial Spillover Effect
Based on population migration data from Baidu and the spatial Durbin model, this paper examines the impact of population mobility on the spatial transmission of COVID-19 and provid...
Fusion of Machine learning for Detection of Rumor and False Information in Social Network
Fusion of Machine learning for Detection of Rumor and False Information in Social Network
In recent years, spreading social media platforms and mobile devices led to more social data, advertisements, political opinions, and celebrity news proliferating fake news. Fake n...
Monitoring of anthropogenic landslide activity with combined UAV and LiDAR-derived DEMs: a case study of the Czerwony Wawoz landslide (SW Poland. Western sudetes)
Monitoring of anthropogenic landslide activity with combined UAV and LiDAR-derived DEMs: a case study of the Czerwony Wawoz landslide (SW Poland. Western sudetes)
The paper presents the results of fieldwork and geodetic surveys carried out on a landslide in Wleń (Western Sudetes, Izerskie Foothills). This reactivated landslide occured firstl...
Convolutional Neural Networks for Image-Based High-Throughput Plant Phenotyping: A Review
Convolutional Neural Networks for Image-Based High-Throughput Plant Phenotyping: A Review
Plant phenotyping has been recognized as a bottleneck for improving the efficiency of breeding programs, understanding plant-environment interactions, and managing agricultural sys...