A Python library for the analysis of rod-like nanostructures from SEM images.
The nanostructure-analysis.py script provides an example of a full analysis of experimental Scanning Electron Microscopy (SEM) images to extract nanostructure data using the code provided in the nanoanalysis package. Further information on the latter can be found in the package readme.
The script can be readily adapted to analyse any appropriate dataset by supplying the paths to the folders storing images, segmentation masks, and results, which are controlled, respectively, by the IMAGE_FOLDER, SEGMENTATION_FOLDER, and RESULTS_FOLDER variables.
Instance segmentation data should be obtained via external software, and supplied in one of the supported formats. Currently, the CVAT for images 1.1 and COCO 1.0 formats are supported.
The code in the repository requires a Python 3.7 or newer installation, which must include Numpy, Pandas, Scipy, Scikit-Image, and Scikit-Learn. Skeleton analysis relies on skan. Optional plotting functions require matplotlib, and optional metadata extraction by OCR relies on PyTesseract.
A complete list of the packages in the virtual environment used for the development of this code is included in environment.yml.