AI-PCC Reporting Template

This work attempts to make a reporting template for MPEG AI PCC to analyze the testing results, such as calculating the BD-Rates, feawing the RD curves, etc.

Requirments

pip install pandas matplotlib

We use the Bjontegaard metric calculation function from: https://github.com/Anserw/Bjontegaard_metric/blob/master/bjontegaard_metric.py Other resources are also available: https://github.com/mauriceqch/pcc_geo_cnn/blob/master/src/metrics.py https://www.mathworks.com/matlabcentral/fileexchange/27798-bjontegaard-metric

Usages

python test.py --csvdir1='csvfiles/reporting_template_lossy.csv' --csvdir2='csvfiles/test.csv' --csvdir_stats='csvfiles/reporting_template_stats.csv' --xlabel='bppGeo' --ylabel='d1T'

python test_mean.py --category='solid' --csvdir1='csvfiles/reporting_template_lossy.csv' --csvdir2='csvfiles/test.csv' --csvdir_stats='csvfiles/reporting_template_stats.csv' --xlabel='bppGeo' --ylabel='d1T'

Results

Template	python script	G-PCC Excel
house_vox12	-11.1	-10.9

Update

20220602: Bitrate on figure x-axis instead of total bits; summary for the average results.

20220512: Test on csv files including more sequences, the csv files are provided by Muhammad.

TODO

gitlab