Scene Classification

Instructions for use

With the continuous improvement in the resolution of satellite and aerial remote sensing images, more useful data and information can be obtained from these images. Depending on the application requirements of different scenarios, the processing of remote sensing images varies. To analyze and manage this data effectively, it is necessary to assign semantic labels to the images based on their content, and scene classification is an important approach to address this type of problem.

Scene classification involves classifying image data based on certain land cover features, distinguishing regions with similar characteristics, and assigning scene labels to each image. Thus, the key to scene classification lies in the extraction of image features.

The results of scene classification consist of uniformly sized (32*32 pixels) vector grids, with each grid carrying the scene label corresponding to its location on the image. This is a coarse-grained classification method, less detailed than pixel-level classification. Scene classification is generally used for large-scale classification tasks, such as local climate zone classification in urban areas.

Parameter Description

Parameter name Default Parameter description Parameter Type
File Type Dataset Specify File Type. Select Dataset or Folder Type. Boolean
Model File   Specify the Model File (*.sdm) to use for Scene Classification String
Amount of single step operation 1 Specifies the number of pictures to process at one time when reasoning Integer
Processor type GPU Specifies the processor type String
GPU number 0 Input the GPU number to support multi-card reasoning, that is, input multiple GPU numbers. If "0, 1, 2" is entered, it will automatically become "012" String
Other Parameter Settings
(Optional)
false Check this box to set the Bounds Dataset/Datasource parameter. Boolean
Result Datasource   Datasource where the specified Dataset storing the result resides Datasource
Result Dataset SceneClassificationResult The specified Resulting dataset name. String

Output Result

Parameter name Parameter description Parameter Type
Result Dataset Result Dataset DatasetVector