Hi there!
First of all, I want to express my sincere appreciation for your excellent work on SceneFun3D — the dataset and its novel interaction annotations have brought great value to the 3D indoor scene research community. It’s truly impressive to see such a comprehensive effort on high-resolution real-world scene understanding!
I’ve been exploring the released dataset recently and noticed that the current split includes 200 training samples, 30 validation samples, and 85 test samples.
However, the paper mentions that SceneFun3D contains more than 14.8k highly accurate interaction annotations across 710 3D indoor scenes.
I’m really looking forward to leveraging the full version of the dataset for my research, and would also be grateful to access the model code corresponding to the paper if possible.
Could you kindly share any updates on the release timeline for the complete dataset and the associated code?
Thank you again for your remarkable contribution, and I appreciate your time and response in advance!
Hi there!
First of all, I want to express my sincere appreciation for your excellent work on SceneFun3D — the dataset and its novel interaction annotations have brought great value to the 3D indoor scene research community. It’s truly impressive to see such a comprehensive effort on high-resolution real-world scene understanding!
I’ve been exploring the released dataset recently and noticed that the current split includes 200 training samples, 30 validation samples, and 85 test samples.
However, the paper mentions that SceneFun3D contains more than 14.8k highly accurate interaction annotations across 710 3D indoor scenes.
I’m really looking forward to leveraging the full version of the dataset for my research, and would also be grateful to access the model code corresponding to the paper if possible.
Could you kindly share any updates on the release timeline for the complete dataset and the associated code?
Thank you again for your remarkable contribution, and I appreciate your time and response in advance!