Segmentation Data Selection

Select the segmented image stack to convert into a 3D mesh. Use results from segmentation inference or annotation masks.

Compatible Data Sources

• Segmentation Results: Output from U-Net inference (Recent Results section)

• Annotation Masks: Completed annotations from the Annotation Tool

• Uploaded Files: Any labeled integer TIFF stack

File Requirements

• Format: Multi-page TIFF (.tif, .tiff)

• Pixel Values: Integer class labels (0 = background, non-zero = classes)

• Data Type: An integer type - uint8, uint16, uint32, int8, int16, or int32. The data is used as-is; no bit-depth conversion is applied.

Validation

When you select a file, it is checked before you can continue. A file is rejected if:

• No class labels can be read from it

• Only background is present (no object classes)

• More than 50 unique values are found (this looks like raw image data, not discrete labels)

• The data type is not one of the supported integer types

If a freshly uploaded file fails these checks, it is automatically removed from your workspace so it does not clutter the file list.

Class Detection

The system automatically detects the unique class values in your data. In the next step you can generate meshes for all classes at once or for a single class.

Note: Background pixels (value 0) are excluded from mesh generation. Only non-zero class values produce surfaces.