A method for simulating shielding shadows to enhance plant data samples
A technology of data samples and shadows, applied in character and pattern recognition, biological neural network models, instruments, etc., can solve the problems of insufficient classification data samples, noise, blur, etc.
Inactive Publication Date: 2019-06-04
GUANGXI TALENTCLOUD INFORMATION TECH
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
In deep learning, when collecting data samples for training, some classified data samples are seriously insufficient, and the data set is too small to easily cause the model to overfit.
Data samples are usually rotated, translated, flipped, scale transformed, based on image saturation and contrast changes, noise, blur, etc. to expand and enhance data samples; however, the above methods of enhancing data samples are difficult to meet the complex plant growth environment
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0037] The solution of the present invention will be described in detail below with reference to the accompanying drawings.
[0038] A method for simulating occlusion shadows to enhance plant data samples, taking citrus leaves as an example, considering that citrus leaves often occlude each other and produce occlusion shadows, the method of simulating occlusion shadows enhances data samples. Including the following steps:
[0039] S0: Establish a network model;
[0040] S0-1: Take several photos containing different citrus leaves as the citrus leaf sample library, the number of pictures shall not be less than 5000;
[0041] S0-2: Use the labelme labeling tool to label the citrus leaves in the picture obtained by S0-1, and frame the citrus leaves in the picture;
[0042] S0-3: Preprocess the picture marked in S0-2, use OpenCV's RotateImage to rotate the picture, use LightImage to process the picture brightness, and realize data enhancement; use caffe's Scale layer and BatchNo...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention belongs to the field of agricultural product image recognition and intelligent computing, and particularly relates to a method for simulating shielding shadows to enhance plant data samples. The invention relates to a method for simulating shielding shadows to enhance plant data samples, which comprises the following steps: selecting a plurality of most typical shadow templates of aplant, and constructing a standard shadow template library; Inputting a picture, identifying the picture by the network model, and returning to detect a target needing shadow increase in the picture;Randomly selecting any shadow template from the standard shadow template library; Adjusting the size of the selected shadow template; Moving the shadow template of which the size is adjusted in the input picture, and enabling the shadow template to be intersected with an identification target of the input picture; And according to the shadow intensity of the target, carrying out brightness adjustment on an area where the shadow template intersects with the recognition target of the input picture to obtain a new data sample. According to the method, the data sample is enhanced by adopting a method of simulating shielding shadows, the method is closer to the actual scene of plant growth, and another method is provided for enhancing the data sample.
Description
technical field [0001] The invention belongs to the field of image recognition and intelligent calculation of agricultural products, and in particular relates to a method for simulating occlusion shadows and enhancing plant data samples. technical background [0002] In the field of agriculture, it is difficult to collect plant data samples, especially some uncommon samples are even more difficult to collect. In deep learning, when collecting data samples for training, some classification data samples are seriously insufficient, and the data set is too small to easily cause the model to overfit. Data samples are usually rotated, translated, flipped, scale transformed, based on image saturation and contrast changes, noise, blur, etc. to expand and enhance data samples; however, the above methods of enhancing data samples are difficult to meet the complex plant growth environment . Contents of the invention [0003] According to the above-mentioned technical problems, the ...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04
Inventor 韦光亮王筱东吴光杰苏世宁张玉国龚骏逸黄彬
Owner GUANGXI TALENTCLOUD INFORMATION TECH