In the Training Data Cleaning tab, under the Training Data Management group, two core functions, Add Image and Export Training Data, are organized to support users in dynamically expanding and outputting sample data in the cleaning project, forming a complete data processing closed loop.
Add Image
Instructions for use
It is used to batch import new image files into the current training data cleaning project to expand the sample quantity. The system automatically performs standardized processing on the newly added images (such as renaming and size verification) and integrates them into the project's management system, facilitating subsequent classification and editing.
When adding images, the following limitations need to be noted:
- Image specifications: Currently, only *.jpg format image files are supported for addition, and the size and number of bands of the new images must be consistent with the original training data used when creating the project.
- Storage rules for adding images: To avoid naming conflicts, the system automatically adds the prefix "n_" to the names of newly added image files (e.g., "1_" for the first addition, "2_" for the second addition, and so on), and creates corresponding numbered folders under the project's Storage path (e.g., "1" for the first addition, "2" for the second addition, and so on) to store these processed image files.
Available starting from the SuperMap iDesktopX 2026 version.
Operation Instructions
- In the Training Data Cleaning tab -> Training Data Management group, click the Add Image button to pop up the Add Image dialog box.
- In the dialog box, manage the images to be added using the toolbar above the file list:
- Add: Add single or multiple selected image files.
- Add Folder: Add all images in the selected folder that meet the format. You can check the Search files in subfolders option for recursive addition.
- Select All: Select all records in the list.
- Select Reverse: Inverse selection of all records in the list.
- Delete: Delete the selected records from the list.
- After confirming the file list is correct, click the OK button at the bottom of the dialog box. The system will verify and process the images. Upon successful processing, the newly added images will be integrated into the current project.
Export Training Data
Instructions for use
It is used to export high-quality sample data that has been cleaned, classified, and edited in the project to a local directory according to user-specified categories. The exported data maintains a standard training data organizational structure and can be directly used for subsequent AI model training tasks.
Available starting from the SuperMap iDesktopX 2026 version.
Operation Instructions
- In the Training Data Cleaning tab -> Training Data Management group, click the Export Training Data button to pop up the Export Training Data dialog box.
- In the dialog box, set the following parameters:
- Export Directory: Specify the local file directory where the exported data will be stored.
- Data Name: Specify the folder name created under the export directory to identify the training data exported this time.
- Export Mode: When the data name already exists in the destination folder, choose how to handle files with the same name:
- Replace: Directly replace the existing files. Suitable for overwriting old exports with the final version.
- Skip: Keep the existing files and only export new files. Suitable for incremental exports to save time.
- Export After Clearing: First delete all contents in the destination folder, then re-export. Suitable for export scenarios that require absolute purity and no historical mixed data.
- Sample Type: Select the sample category to export.
- Click OK, and the system will export the images and labels of the selected categories in standard training data format.
Related Topics