Guillaume Lavoué Homepage
Contact
glavoue@liris.cnrs.fr
Tél.: (33) 04 72 43 71 36

Subjective quality assessment of 3D models


LIRIS Localized Geometric Artefacts Database


We provide here the dataset associated with the reference below:

Jinjiang Guo, Vincent Vidal, Atilla Baskurt & Guillaume Lavoué, Evaluating the local visibility of geometric artifacts , ACM Symposium on Applied Perception  (SAP) 2015.

11 test models were generated from 4 reference objects. Different distortions were applied (watermarking, simplification, noise, smoothing, quantization). In the experiment, we asked the observers to mark vertices of the distorted models where they perceived noticeable differences as compared with the reference ones (using a brush painting interface). 20 observer took part to the experiment. This experiment provided localized subjective distortion maps. The analysis is available in the paper and supplementary material.

Download the LIRIS Localized Geometric Artefacts Database (32 MB zip file)

LIRIS/EPFL General-Purpose database


This Mean Opinion Score database was created at EPFL and experiments were conducted at EPFL and LIRIS, Université de Lyon. It you use it, please cite: 

Lavoue G, Drelie Gelasca E, Dupont F, Baskurt A, Ebrahimi T. Perceptually driven 3D distance metrics with application to watermarking. In: Proceedings of SPIE.Vol 6312. SPIE; 2006:63120L-63120L-12.

  • 88 models between 40K and 50K vertices were generated from 4 reference objects. Two types of distortion (noise addition and smoothing) were applied with different strengths and at four locations: on the whole model, on smooth areas, on rough areas and on intermediate areas.
  • Subjective evaluations were made at normal viewing distance, using a SSIS (Single Stimulus Impairment Scale) method with 12 observers. 
  • A Microsoft excel document giving all subjective quality scores is included in the above archive. It contains also Mean Opinion Scores after normalization and outlier removal. An other Microsoft excel document provides the objective scores of several recent perceptual metrics from the state of the art.
Download the LIRIS/EPFL General-Purpose database (90 MB zip file)

LIRIS Masking database


The Mean Opinion Score database was created at LIRIS, Université de Lyon. It you use it, please cite: 

Lavoué G. A local roughness measure for 3D meshes and its application to visual masking. ACM Transactions on Applied Perception (TAP). 2009;5(4).

  • 26 models between 9K and 40K vertices were generated from 4 reference objects. The only distortion is noise addition applied with three strengths, either on smooth or rough regions.
  • Subjective evaluations were made at normal viewing distance, using a Multiple Stimulus Impairment Scale method with 11 observers.
  • A Microsoft excel document giving all subjective quality scores is included in the above archive. It contains also Mean Opinion Scores after normalization and outlier removal. An other Microsoft excel document provides the objective scores of several recent perceptual metrics from the state of the art. 
Download the LIRIS Masking database (18 MB zip file)


Notice:
The Armadillo model from these databases is a manifold/simplified version of the original model that was created from scanning data by the Stanford Computer Graphics Laboratory.

The Dinosaur and Igea models are the courtesies of the Cyberware Inc.
The Bimba, RockerArm and vaseLion models are the courtesies of the AIM@SHAPE project.


3D Segmentation benchmark


The goal of this 3D-mesh segmentation benchmark is to provide an automatic tool to evaluate, analyse, and compare the different automatic 3D-mesh segmentation algorithms. It provides a corpus of segmented 3D models, an easy online evaluation tool and some comparison results for recent algorithms. It is available here.

Reference:  Halim Benhabiles, Jean-Philippe Vandeborre, Guillaume Lavoué and Mohamed Daoudi, A comparative study of existing metrics for 3D-mesh segmentation evaluation, The Visual Computer, Vol. 26, No. 12, pp. 1451–1466, 2010.

3D Mesh Watermarking benchmark

The proposed 3D mesh watermarking benchmark has three different components: a data set, a software tool and two evaluation protocols. The data set contains several "standard" mesh models on which we suggest to test the watermarking algorithms. The software tool integrates both geometric and perceptual measurements of the distortion induced by watermark embedding, and also the implementation of a variety of attacks on watermarked meshes. Besides, two different application-oriented evaluation protocols are proposed, which define the main steps to follow when conducting the evaluation experiments. This benchmark is available here.

Reference:  Kai Wang, Guillaume Lavoué, Florence Denis, Atilla Baskurt and Xiyan He, A Benchmark for 3D Mesh Watermarking, IEEE Shape Modeling International (SMI) , Avignon, France, June 2010.