3D-to-2D Image Segmentation Dataset Generator (Automated mask creation)
Developed a 3D-to-2D dataset generator to create segmentation training data by rendering multi-angle 2D images from Blender 3D models. Implemented automated background removal and contour-based segmentation to generate high-quality segmentation masks without manual annotation. Produced masks suitable for training computer vision models that require pixel-level class delineation. • Rendered 3D models into multi-angle 2D images using Blender • Applied background removal to isolate foreground regions • Used contour-based methods to derive segmentation masks • Generated segmentation outputs for downstream CV model training