---------------------------------------------------------------------------------------------
BASICS: Broad Quality Assessment of Static Point Clouds in a Compression Scenario
---------------------------------------------------------------------------------

For a detailed explanation of the dataset and the analysis done, please refer to the related publication:

@ARTICLE{10403987,
  author={Ak, Ali and Zerman, Emin and Quach, Maurice and Chetouani, Aladine and Smolic, Aljosa and Valenzise, Giuseppe and Le Callet, Patrick},
  journal={IEEE Transactions on Multimedia}, 
  title={BASICS: Broad Quality Assessment of Static Point Clouds in a Compression Scenario}, 
  year={2024},
  pages={1-13},
  doi={10.1109/TMM.2024.3355642}}



BASICS paper is accepted to be published in IEEE TMM. The above article will be updated later on with the correct DOI.

If you use the dataset in your research, please cite the paper above.

This dataset contains 75 source point clouds (SRC).
Each SRC is compressed with 4 different compression algorithms (geocnn, gpcc-predlift, gpcc-raht, vpcc)
Subjective experiment is conducted in 60 separate sessions which are referred as playlists in the dataset repository. 
Each playlist was rated by ~60 unique observers. You can find the individual opinion scores for each playlists in the corresponding folder. 
IMPORTANT: Check below for the explanation regarding the individual opinion scores.


Organization of the dataset
---------------------------
+ "PointClouds" folder contains 2 folders:
	+ "SRC": contains the 75 source point clouds
	+ "PPC": contains the 1494 compressed point clouds

+ "VideoRenderings" folder contains 2 folders:
	+ "SRC": contains the video renderings of the SRCs that were used in the subjective experiment
	+ "PPC": contains the video renderings of the PPCs that were used in the subjective experiment

- src_info.csv file contains the basic information regarding to the SRCs. Columns corresponds to:
	"name": Name of the SRC
	"Split": Training/Validation/Test split information in the ICIP 2023 PCQVA Grand challenge
	"PointSize": Point size used for rendering, see the paper for more detail.
	"NbPoints": Number of points in the SRC. 
	"Capture": Method used to capture the point cloud
	"SemCat": One of the three semantic categories which the point cloud belongs to 
	"Content": Brief description of the point cloud content
	"Creator": Creator alias or name
	"Copyright": Copyright of the source model
	"SketchFab": Link to the SketchFab page of the model (if downloaded from SketchFab)

- "ppc_bitperpoints.csv" file contains the BitPerPoint(BPP) information of each PPC. 
    (note that, no BPP information is given for GeoCNN compressed point clouds. See the paper for the details.)

- "metric_predictions.csv" file contains all the metric predictions that have been benchmarked in the BASICS journal paper.

+ "SubjectiveOpinions" folder contains several files about the Individual and Mean Opinion Scores collected in the experiment as well as the partition of each content in the ICIP 2023 Grand Challenge (Training, Validation, Test).
	- "MOS_CI_all.csv" file contains the Mean Opinion Scores (MOS) and 95% Confidence Intervals(CI) for all the PPCs in the dataset. 
	- "IndividualScores_all.csv" file contains the individual opinion scores for each PPC. Note that, the observer numbers in the columns does not correspond to the same observers, for this information please see below,
	+ "votes_per_playlist" folder contains 60 *.csv file with individual opinion scores of ~25 PPC.
		(Each column in these csv files is a unique observer. Therefore, if you plan to analyse the agreement/reliability of the observers, you must use these individual playlists files, rather than IndividualScores_all.csv file.)


Questions
---------

For any questions and problems regarding the dataset, please contact to:

Ali Ak, ali.ak@univ-nantes.fr, ali7ak@gmail.com
Emin Zerman, emin.zerman@miun.se
---------------------------------------------------------------------------------------------