A multi-class target detection method in remote sensing images based on sample reweighting
A remote sensing image and target detection technology, applied in the field of remote sensing image processing, can solve problems such as large differences in the aspect ratio of target samples, and achieve the effects of improving detection accuracy, fast inference speed, and optimizing training models.
Active Publication Date: 2022-04-26
RES & DEV INST OF NORTHWESTERN POLYTECHNICAL UNIV IN SHENZHEN +1
View PDF7 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] Aiming at the problem of large difference between the salient feature extraction and the aspect ratio of target samples in the target detection task of optical remote sensing images based on deep learning technology, the present invention proposes a multi-class target detection method for remote sensing images based on sample reweighting
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
[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments, and the present invention includes but not limited to the following embodiments.
[0046] Such as figure 1 As shown, the present invention provides a multi-class target detection method in remote sensing images based on sample reweighting, and constructs a new target detection network. For better illustrating the present invention, present embodiment is in hardware environment: Intel (R) Core (TM) i3-8100 CPU computer, 8.0GB memory, graphics card model: Titan X (Pascal), usable memory is 12GB, software environment : Experiment under Pycharm2016 and Ubuntu 16.04.5LTS. The experiment uses the public optical remote sensing database DIOR. There are 23,463 images in the data set, and a total of 192,472 horizontal box instances are labeled for 20 categories, and the pixels of each image are 800×800. In order to verify the rationality and effectiveness of the ...
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 provides a multi-category target detection method in remote sensing images based on sample reweighting. First, image data augmentation processing and scaling preprocessing are performed; then, a target detection network is constructed, including a feature extraction module, a feature enhancement module, and a detection head module. In order to realize the salient expression of features, feature enhancement is performed on some feature levels operation; then, the end-to-end training process of the network is carried out, and the sample reweighting strategy is used to guide the training network to pay more attention to the target samples with large aspect ratio differences, so as to optimize the training model; finally, to realize the target detection process, the remote sensing The image is input into the trained target detection network to obtain the category prediction value and its coordinate offset of each prior box, and then use non-maximum value suppression to filter out the detection results with high overlap rate for the same target. The invention has higher detection accuracy and speed of the remote sensing image target.
Description
technical field [0001] The invention belongs to the technical field of remote sensing image processing, and in particular relates to a method for detecting multiple types of targets in remote sensing images based on sample reweighting, which can be used to improve the detection effect of target categories with large aspect ratio differences in remote sensing image data sets. Background technique [0002] Remote sensing image target detection is a key technology in the application field of remote sensing big data information. The close combination of high-resolution remote sensing image data and geographic information systems will be used in future urban road planning, engineering project evaluation, and monitoring and evaluation of renewable resources. There will be broad prospects for development. With the advent of the era of big data and the substantial improvement of computer hardware performance, the target detection algorithm based on deep learning technology has broke...
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 Patents(China)
IPC IPC(8): G06V20/13G06V10/80G06V10/774G06V10/764G06V10/776G06V10/778G06V10/82G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V20/13G06V2201/07G06N3/045G06F18/217G06F18/24G06F18/253G06F18/214
Inventor 程塨司永洁姚西文韩军伟郭雷
Owner RES & DEV INST OF NORTHWESTERN POLYTECHNICAL UNIV IN SHENZHEN