• Scene variation
• Documentation variation
• Segmentation and metadata
Objects
Fully procedural tax documentation with synthetic information for training Optical Character Recognition (OCR) models. The system varies paper materials, creases, textures, backgrounds, layouts, fonts, and colors to create diverse training data.