nanodevice_flakedetect_combine

To combine per-material detections into a unified coordinate system, apply per-material coordinate transforms like inverted warp for bottom layers and direct warp for top layers. This extracts contours and outputs a unified traces.json file with material metadata and area measurements.

To generate overlay images from nanodevice detection data, run the transform script after detection and alignment steps. This produces overlay-ready imagery alongside a combine_report.json containing transform summaries and per-material diagnostics for visualization.

Yes, you need OpenCV, NumPy, and Shapely to transform nanodevice traces. These dependencies are required to execute per-material coordinate transforms, extract contours, perform footprint clipping, and generate the unified traces.json output.

The best way to extract contours and build a unified traces.json is to apply per-material coordinate transforms into a full_stack system. This approach handles layer-specific clipping like footprints for graphene and pass-throughs for hBN, appending material metadata and area measurements.

Misaligned coordinates in traces.json occur when per-material detections are not properly transformed into the full_stack coordinate system. Ensure you apply the inverted warp for bottom_part layers, direct warp for top_part, and correct footprint clipping for materials like graphene.