Instructions for Use
With the continuous improvement in the resolution of satellite remote sensing images and aerial remote sensing images, more useful data and information can be obtained from them. For applications in different scenarios, there are varying requirements for processing remote sensing images. To effectively analyze and manage these remote sensing image data, it is necessary to assign semantic labels to the images based on their content. Scene classification is an important approach to solving this type of problem.
Scene classification categorizes image data based on specific ground object features, distinguishing areas with similar characteristics and assigning scene labels to each image. Therefore, the key to scene classification lies in the extraction of image features.
The scene classification result is a vector grid of uniform size, where each grid cell carries the scene label of the imagery at its corresponding location. It is a coarse-grained classification method at a level lower than the pixel. Scene classification is generally used for large-scale classification tasks, such as classifying local climate zones in cities.
Parameter Description
-
Source Dataset: Select the imagery that requires scene classification. It supports two types: Dataset and Folder.
-
-
If the File Type is set to Dataset, it supports input file-based image data (e.g., *.tif, *.img, etc.), image datasets, and mosaic datasets.
-
If the File Type is set to Folder, you need to input the image folder path.
-
Notes:- Only supports images in 8-bit unsigned format.
- Supports three-band and multi-band imagery (imagery with four or more bands). The band order of the imagery must be consistent with the sample images used during model training.
-
Model File: Select the model file (*.sdm).
-
Batch Size: Due to the large volume of remote sensing image data, and limited by computer performance, the model cannot load all data at once. Therefore, a tile-based reading and processing method is used during inference. Batch size refers to the number of tiles processed simultaneously in one inference step. Appropriately increasing the batch size can improve interpretation efficiency, but it is limited by the video memory (for GPU inference) or system memory (for CPU inference) of the inference device.
-
Processor Type: You can use the computer's Central Processing Unit (CPU) or Graphics Processing Unit (GPU) to process data.
-
GPU ID: Specify the GPU ID for data processing. The default is 0. If "Processor Type" is set to CPU, this value is fixed at -1. If using GPU, you need to query the GPU ID via the "nvidia-smi" command in the system command line. The GPU ID should be based on the query result. If there are multiple GPUs, you can specify the GPU ID(s) used for processing. To use multiple GPUs, separate their IDs with English commas ",", e.g., "0,2,3" indicates using GPUs with IDs 0, 2, and 3.
-
Other Parameters: Check "Other Parameters" to input a region dataset that has the same coordinate system as the source data, used to limit the inference extent.
-
Result Data: This is the result vector dataset for scene classification. Set the datasource and name for saving the result dataset. For batch inference, the original image name will be appended to the vector dataset name using an underscore "_" to distinguish the object extraction results from multiple images.
Related Topics