Submitted anonymously.

Submission data

Full nameSQN: Weakly-Supervised Semantic Segmentation of Large-Scale 3D Point Clouds
DescriptionSemantic Query Network trained with only 0.1% randomly selected labels
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)Tensorflow
HardwareIntel Coreā„¢ i9-10900X CPU, NVIDIARTX Titan GPU
Submission creation date24 Mar, 2021
Last edited25 Jan, 2022

3D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
0.5690.6760.6960.6570.4970.7790.4240.5480.5150.3760.9020.4220.3570.3790.4560.5960.6590.5440.6850.6650.556