Citrus green fruit detection via improved feature network extraction
Jianqiang Lu1,2,3, Ruifan Yang1,3, Chaoran Yu4,5
1College of Electronic Engineering (College of Artificial Intelligence), South China Agricultural University, Guangzhou, China.
Frontiers in Plant Science
|December 29, 2022
Summary
This study introduces an improved Mask-RCNN model for accurate citrus green fruit detection, enhancing yield prediction and management. The new method achieves higher accuracy by fusing deep and shallow features, aiding intelligent citrus production.
Area of Science:
- Agricultural Science
- Computer Vision
- Machine Learning
Background:
- Accurate citrus green fruit identification is vital for yield optimization but challenging due to similar fruit and background colors, impacting segmentation accuracy.
- Current deep learning methods achieve 88% accuracy, meeting basic application needs, but improvements are sought for better performance.
Purpose of the Study:
- To develop an improved Mask-RCNN model for enhanced citrus green fruit detection.
- To address the challenge of poor segmentation accuracy caused by the visual similarity between citrus green fruits and their background.
Main Methods:
- Implemented an improved Mask-RCNN model by fusing deep and shallow features using ResNet backbone.
- Introduced a combined connection block to reduce channel numbers and enhance model accuracy.
- Collected and utilized a dedicated citrus green fruit image dataset for testing and comparison.
Main Results:
- The improved Mask-RCNN model achieved an average detection accuracy of 95.36%, a 1.42% increase over the standard Mask-RCNN.
- The area under the precision-recall curve increased to 0.9673, a 0.3% improvement.
Conclusions:
- The enhanced Mask-RCNN model significantly improves citrus green fruit detection accuracy.
- This method effectively reduces background interference, offering a valuable tool for intelligent citrus production.
Related Concept Videos
02:16Light Acquisition
8.6K
In order to produce glucose, plants need to capture sufficient light energy. Many modern plants have evolved leaves specialized for light acquisition. Leaves can be only millimeters in width or tens of meters wide, depending on the environment. Due to competition for sunlight, evolution has driven the evolution of increasingly larger leaves and taller plants, to avoid shading by their neighbors with contaminant elaboration of root architecture and mechanisms to transport water and nutrients.
8.6K
00:56Extraction: Advanced Methods
506
Metal ions can be separated from one another by complexation with organic ligands–the chelating agent– to form uncharged chelates. Here, the chelating agent must contain hydrophobic groups and behave as a weak acid, losing a proton to bind with the metal. Since most organic ligands used in this process are insoluble or undergo oxidation in the aqueous phase, the chelating agent is initially added to the organic phase and extracted into the aqueous phase. The metal-ligand complex is...
506
01:03IR Frequency Region: Fingerprint Region
1.0K
IR spectra are divided into two main regions: the diagnostic region and the fingerprint region. The diagnostic region of the spectrum lies above 1500 cm−1. The absorptions resulting from single-bond vibrations of the N–H, C–H, and O–H stretch at higher wavenumbers and appear on the left side of the spectrum. The stretching absorptions of the C≡C and C≡N occur between 2100–2300 cm−1. In contrast, those arising from stretching absorptions of the...
1.0K