Planned intervention: On Thursday 19/09 between 05:30-06:30 (UTC), Zenodo will be unavailable because of a scheduled upgrade in our storage cluster.
Published August 9, 2023 | Version v1
Dataset Open

3D-rhi-synth-2000- Synthetic Rhinophyma Visual Dataset

  • 1. Dublin City University

Description

In the real world, only a handful of data is available for the Rhinophyma skin condition, typically numbering in the hundreds. This repository contains a Synthetic Dataset of Rhinophyma, generated through 3D head models of one male and one female. The purpose of this data generation is to address the data scarcity of the Rhinophyma skin condition within the medical visual data and computer vision community. By generating such data, we aim to bridge the gap in data scarcity for this disease condition, as well as introduce a proof-of-concept methodology for generating synthetic data for specialized disease conditions.

The highlight folder 'highlight_female_male_rendered' provides a glimpse of the entire dataset. The file '2000_deformations.npy' contains the 2000 values of deformations applied during rendering.

The dataset is divided into two main folders: 'female_rendered' and 'male_rendered'. Within each of these folders, there are three subfolders: 'configu', 'images', and 'points.

1. 'configu': This subfolder contains `.json` files with configuration details for each model. The files include various parameters, such as:
   - "total_num_cameras": the total number of cameras.
   - "active_camera_name": the name of the active camera.
   - "camera_focal_len": the camera's focal length.
   - "camera_loc": the camera's location.
   - "camera_rot": the camera's rotation.
   - "nose_deformation_severity": a measure of the severity of nose deformation.
   - "label": the label for the model (e.g., "Severe").
   - "nose_variants": additional details about nose variants.

2. 'images': This subfolder contains the rendered images in resolution 960x540. They are named according to the following convention e.g.'Nose_Deformation_Severity_0_2.716669764843742_Camera_00001', with specific details related to the deformation severity and camera number. There are images for 10 different cameras.

3. 'points': This subfolder contains polygon files corresponding to each model. These files represent the deformations applied to the models during rendering.

|-- Dataset Root
    |-- female_rendered
    |   |-- configu
    |   |   |-- Nose_Deformation_Severity_0_2.212930927821943_Camera_00001.json
    |   |   |-- Nose_Deformation_Severity_0_2.212930927821943_Camera_00002.json
    |   |   |-- ...
    |   |
    |   |-- images
    |   |   |-- Nose_Deformation_Severity_0_2.212930927821943_Camera_00001.png
    |   |   |-- Nose_Deformation_Severity_0_2.212930927821943_Camera_00002.png
    |   |   |-- ...
    |   |
    |   |-- points
    |   |   |-- Nose_Deformation_Severity_0_2.212930927821943.ply
    |   |   |-- ...
    |   
    |-- male_rendered
        |-- configu
        |   |-- Nose_Deformation_Severity_0_2.716669764843742_Camera_00001.json
        |   |-- Nose_Deformation_Severity_0_2.716669764843742_Camera_00002.json
        |   |-- ...
        |
        |-- images
        |   |-- Nose_Deformation_Severity_0_2.716669764843742_Camera_00001.jpg
        |   |-- Nose_Deformation_Severity_0_2.716669764843742_Camera_00002.jpg
        |   |-- ...
        |
        |-- points
            |-- Nose_Deformation_Severity_0_2.716669764843742.ply
            |-- ...

Together, these folders and files comprise a dataset designed to represent and analyze the Rhinophyma condition in both male and female 3D head models that we have created. These 3D models will be made available upon a genuine request to the authors of this dataset.

Files

female_rendered.zip

Files (25.3 GB)

Name Size Download all
md5:06d146bc93dcb68bc8e03bc6df677ae6
16.1 kB Download
md5:6f82706a816279f860eecbaa7b355649
12.7 GB Preview Download
md5:efcb778fe9d7b1c06e3ea1c84b5dca4c
25.3 MB Preview Download
md5:7919320be3e8f7964126c5e6e433c875
12.6 GB Preview Download

Additional details

Funding

SFI Centre for Research Training in Digitally Enhanced Reality 18/CRT/6224
Science Foundation Ireland

Subjects

Computer Vision
Medical/Clinical Image Analysis
Deep Learning
Synthetic Data/Generation