Synthetic Image Assisted Deep Learning Framework for Detecting Defects During Composite Sheet Layup

The repository consists of the codebase and dataset for work presented at ASME's IDETC-CIE Conference, 2022

The link to the entire annotated dataset can be found here. In case there are any issues accessing the dataset kindly reach out via email to: [email protected] or [email protected]

The codebase consists of two subfolders:

Synthetic Image Generation

The Synthetic Image Generation module contains the Texture File of Carbon Fiber, Procedure to Generate your own texture and also talks about generation of synthetic images The code, blender file and procedure to generate synthetic images can be found here

ResNeSt-based Image Segmentation Model

The image segmentation module is based on the ResNest Framework The code and implementation details are discussed here