Open pureexe opened 10 months ago
Sorry, I am very busy these days. I am afraid that I have no time to run the code and check the differences. One thing I think you might miss is that I removed the black regions and warped the panoramas to some extent such that the object is placed at a new center. This is similar to the method Gardner et al. Learning to Predict Indoor Illumination from a Single Image, SIGGRAPH Asia, 2017..
You might keep the same setting for fair comparison when compared with other methods, since I found lots of hyperparameters would affect the final results.
I would like to thank you for creating such a great work.
I'm currently working on light estimation and want to compare it with the Stylelight.
However, when I ran your code on the Laval indoor dataset, the score is quite different. So, I would like to ask if this is still acceptable. If not, I am looking for the correct way to run an evaluation.
Let me specify how I got this score so you can point out which step I did wrong.
data_prepare_laval.py
root_path
intest_lighting.py
to the corrected pathtest
usingevaluation/tonemap.py
evaluation/tonemap.py
evaluation/test_render.sh
evaluation/test_render.sh
evaluation/test_rmse.sh --fake <stylelight's tone-mapped dir> --real <ground truth's tone-mapped dir>
Best regard.