Object Extraction
Feature Description
Object extraction uses neural network models to pre-select regions of interest in imagery, then classifies pixels within these regions to obtain object boundary information, supporting object-level spatial analysis.
Parameter Description
| Parameter Name | Default Value | Description | Type |
|---|---|---|---|
| File Type | Dataset | Specifies the file type selection between dataset or folder format. | Boolean |
| Model File | Specifies the model file (*.sdm) for object extraction. | String | |
| Probability Threshold | 0.5 | The system calculates probability scores for detected objects matching target features. Only objects with scores above this threshold will be saved in detection results. Default: 0.5. | Double |
| Deduplicate Threshold | 0.3 | Specifies the threshold for removing duplicate detections. | Double |
| Batch Size | 1 | Specifies the number of images processed per inference step. | Integer |
| Processor Type | GPU | Specifies the processor type for computation. | String |
| GPU ID | 0 | Enter GPU IDs, supporting multi-GPU inference by entering multiple IDs. E.g., "0,1,2" will become "012". | String |
| Other Parameter Settings (Optional) |
false | Check this box to enable bounds dataset/datasource parameter settings. | Boolean |
| Result Datasource | Specifies the datasource where the resulting dataset will be stored. | Datasource | |
| Result Dataset | objextract_result | Specifies the name for the resulting dataset. | String |
Output
| Parameter Name | Description | Type |
|---|---|---|
| Result Dataset | Resulting dataset | DatasetVector |