Dataset

Broad Quality Assessment of Static Point Clouds (BASICS) dataset will be used in the challenge.  BASICS dataset comprises 75 unique static point cloud from 3 different semantic categories (Humans & Animals, Inanimate Objects, Building & Landscapes).  Each point cloud compressed with 4 different compression algorithms (VPCC, GPCC-RAHT, GPCC-Predlift and GEOCNN) at varying compression levels, resulting in 1494 processed point clouds

Each point cloud is rendered as a 10 seconds video sequence and shown to the participants. The user votes are collected via crowdsourcing, and each participant attended the experiment in their personal spaces. 60 unique observers rated each stimulus.

The dataset is split into three parts as training (45 SRC), validation (15 SRC) and test set (15 SRC). Reference and distorted point clouds (with corresponding MOS, std and ci values) in training set will be publicly available at the development phase.  Reference and distorted point clouds (without MOS, std and ci values) in validation set will be publicly available at the development phase. Test set content will not be publicly available until the end of the challenge.

Please check the "Tracks and Evaluation Criteria" page 

to find the links to the CodaLab pages.