DIO: Dataset of 3D Mesh Models of Indoor Objects for Robotics and Computer Vision Applications

TL;DR

The paper introduces the DIO dataset, a collection of textured 3D meshes of indoor objects for robotics and computer vision, produced via two complementary pipelines: photogrammetry with DSLR/phone imagery and active 3D scanning with a Structure Sensor. It documents image acquisition, color calibration, reconstruction, and post-processing steps to yield high-fidelity meshes, plus a Gazebo-based demonstration to show practical integration into simulation environments. The dataset comprises 141 objects across 13 categories, supported by 3,584 photogrammetry-derived images, and is released with corresponding models and imagery to support perception, manipulation, and neural rendering research. This work enables more realistic simulations and training data for object recognition and planning, particularly benefiting neural rendering and robotics tasks requiring both geometry and texture fidelity.

Abstract

The creation of accurate virtual models of real-world objects is imperative to robotic simulations and applications such as computer vision, artificial intelligence, and machine learning. This paper documents the different methods employed for generating a database of mesh models of real-world objects. These methods address the tedious and time-intensive process of manually generating the models using CAD software. Essentially, DSLR/phone cameras were employed to acquire images of target objects. These images were processed using a photogrammetry software known as Meshroom to generate a dense surface reconstruction of the scene. The result produced by Meshroom was edited and simplified using MeshLab, a mesh-editing software to produce the final model. Based on the obtained models, this process was effective in modelling the geometry and texture of real-world objects with high fidelity. An active 3D scanner was also utilized to accelerate the process for large objects. All generated models and captured images are made available on the website of the project.

Figures (8)

• Figure 1: Sample meshes available in the dataset.
• Figure 2: The process of moving the camera and tripod globally around the target object for image acquisition.
• Figure 5: Editing the raw mesh model of a shell using MeshLab.
• Figure 6: The Structure Sensor scanning process for an office chair.
• Figure 7: Plots demonstrating the distribution of models across categories for the Structure Sensor and Photogrammetry Pipelines.