Orthographic View Dataset Generation

This repository contains scripts for generating multi-view orthographic datasets from 3D models using Blender. Each data item consists of four orthographic views (front, top, back, bottom) with corresponding RGB images, depth maps, and segmentation masks.

Generated Dataset Examples

The following tables showcase sample outputs from the dataset generation scripts. Each table represents a different 3D model rendered from four orthographic viewpoints, with three data types per view: RGB image, segmentation mask, and depth map.

---

Dataset Item 0013

Front	Top	Back	Bottom

Row 1: RGB Images | Row 2: Segmentation Masks | Row 3: Depth Maps

---

Dataset Item 0014

Front	Top	Back	Bottom

Row 1: RGB Images | Row 2: Segmentation Masks | Row 3: Depth Maps

---

Dataset Item 0016

Front	Top	Back	Bottom

Row 1: RGB Images | Row 2: Segmentation Masks | Row 3: Depth Maps

---

Dataset Item 0017

Front	Top	Back	Bottom

Row 1: RGB Images | Row 2: Segmentation Masks | Row 3: Depth Maps

---

Data Structure

Each generated data item contains:

• RGB images - Rendered orthographic views with textures and lighting
• Depth maps - Normalized depth information for each view
• Segmentation masks - Binary masks isolating the object from the background

Four orthographic views per model: