Submitted by Xiaoyang Wu.

Submission data

Full namePoint Transformer V3: Simpler, Faster and Stronger
DescriptionPTv3 trained with Point Prompt Training (PPT)
Publication titlePoint Transformer V3: Simpler, Faster, Stronger
Publication authorsXiaoyang Wu, Li Jiang, Peng-Shuai Wang, Zhijian Liu, Xihui Liu, Yu Qiao, Wanli Ouyang, Tong He, Hengshuang Zhao
Publication venueCVPR 2024
Publication URLhttps://arxiv.org/abs/2312.10035
Input Data TypesUses Color        Uses 3D
Programming language(s)Python
HardwareNVIDIA A100
Websitehttps://github.com/Pointcept/PointTransformerV3
Source code or download URLhttps://github.com/Pointcept/PointTransformerV3
Submission creation date23 Sep, 2023
Last edited2 Mar, 2024

3D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
0.7940.9410.8130.8510.7820.8900.5970.9160.6960.7130.9790.6350.3840.7930.9070.8210.7900.6960.9670.9030.805