Training method and device for cloud segmentation model with high mobility
A segmentation model and high transfer technology, applied in neural learning methods, biological neural network models, image analysis, etc., can solve problems such as affecting observation and transmission, wasting ground system data transmission resources, and losing effective surface information, achieving application prospects. Broad, highly mobile effects
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0058] refer to figure 1 , shows a flow chart of steps of a method for training a cloud segmentation model with high mobility provided by an embodiment of the present invention, as shown in figure 1 As shown, the method may include the following steps:
[0059] Step 101: Obtain training samples and test samples; wherein, the training samples are samples formed by remote sensing images of one load, and the test samples are samples formed by remote sensing images of the above load and other loads.
[0060] The embodiment of the present invention can be applied to a scenario where a remote sensing image of a load is used for model training, and multiple load images are used for model testing to obtain a cloud segmentation model with high mobility.
[0061] The training samples refer to the samples used to train the pre-trained network model.
[0062] The test sample refers to a sample used to test the initial cloud segmentation model trained by using the training sample.
[00...
Embodiment 2
[0130] refer to Figure 5 , which shows a schematic structural diagram of a training device for a cloud segmentation model with high mobility provided by an embodiment of the present invention, as shown in Figure 5 As shown, the device can include the following modules:
[0131] The training sample acquisition module 510 is used to acquire training samples and test samples; wherein, the training samples are samples formed from remote sensing images of a load, and the test samples are formed from remote sensing images of the above load and other loads samples of
[0132] A training data set generating module 520, configured to preprocess the training samples to generate a training data set;
[0133] The initial segmentation model acquisition module 530 is used to train the pre-training network model based on the training data set, and obtain the initial cloud segmentation model according to the accuracy evaluation result;
[0134] The predicted cloud label binary map acquis...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More