nanodevice_flakedetect_detect

To detect and segment vdW materials from microscopy images, automated detectors apply adaptive clustering, multi-threshold sweeps, and region-growing with morphology kernels. This process isolates graphite, graphene, bottom hBN, and top hBN, generating per-material masks, contours, and sidecar JSONs for fabrication pipelines.

Automating flake detection for vdW stack images is best achieved by running independent material detectors across bottom_part and top_part inputs. The system applies adaptive clustering and multi-threshold sweeps to extract per-material masks, contours, and sidecar JSONs for downstream alignment.

Yes, you need OpenCV, scikit-image, numpy, and scikit-learn within a Python 3.x environment to run automated vdW material segmentation. These dependencies provide the required computer vision and machine learning frameworks for adaptive clustering and morphology operations.

Yes, you can extract both top and bottom hBN layers using independent detectors designed for vdW stack images. The system supports per-stack workflows across bottom_part and top_part inputs, generating distinct masks, contours, and sidecar JSONs for each hBN layer.

Graphite and graphene segmentation outputs sidecar JSON files to provide structured metadata for downstream alignment and fabrication pipelines. These JSON files accompany the generated masks and contours, ensuring accurate layer registration and processing within the vdW stack workflow.

Limitations of using adaptive clustering for hBN flake detection include its reliance on multi-threshold sweeps and region-growing with morphology kernels, which require specific image input conditions. The system is specifically tuned for graphite, graphene, bottom hBN, and top hBN, potentially limiting broader material applicability.