ThingiPrint

ThingiPrint is a benchmark dataset that links CAD models to real photographs of their 3D-printed counterparts.
It is intended as an evaluation benchmark for 3D-printed object classification, focusing on generalization to previously unseen object instances.

Associated Paper

Classifying Novel 3D-Printed Objects without Retraining:
Towards Post-Production Automation in Additive Manufacturing

---

Object Overview

The figure below shows example objects from the ThingiPrint dataset.

---

Dataset Contents

ThingiPrint consists of real images of 3D-printed objects corresponding to CAD models sourced from the Thingi10K dataset.

Main Dataset (100 Objects)

• 1000 real images total
• ~10 images per object
• Objects printed using an industrial SLS printer (Sindoh S100)
• Material: white PA12 (polyamide 12)

Auxiliary Dataset (20 Objects)

To evaluate robustness to manufacturing variation, a subset of objects was reprinted using a different printer and material.

• 200 additional real images
• ~10 images per object
• Printer: Prusa MK4
• Material: white PLA

---

Loading the dataset:

from datasets import load_dataset

# Load the Main dataset (100 objects)
ds_main = load_dataset("fanismathioulakis/thingiprint", "main")

# Load the Auxiliary dataset (20 objects)
ds_aux = load_dataset("fanismathioulakis/thingiprint", "auxiliary")

# To access an image from the main set:
sample = ds_main["test"][0]
image = sample["image"]
label = sample["label"]

Data Organization

The dataset follows a folder-per-object structure where each folder name corresponds to the original Thingi10K CAD model ID.

ThingiPrint/
└── test/
    ├── 34785/  <-- Thingi10K Model ID
    │   ├── 01.jpg
    │   └── ...
    ├── 36069/
    └── ...

---

Related Dataset

• CAD models originate from Thingi10K
  https://ten-thousand-models.appspot.com/

---

Citation

@article{mathioulakis2026classifying,
  title={Classifying Novel 3D-Printed Objects without Retraining: Towards Post-Production Automation in Additive Manufacturing},
  author={Mathioulakis, Fanis and Radevski, Gorjan and Cleuren, Silke GC and Janssens, Michel and Das, Brecht and Schauwaert, Koen and Tuytelaars, Tinne},
  journal={arXiv preprint arXiv:2603.07465},
  year={2026}
}