Submitted by Huayao Liu.

Submission data

Full nameUsing dual decoder U-Net and Point Grouping
Publication titleHIDA: Towards Holistic Indoor Understanding for the Visually Impaired via Semantic Instance Segmentation with a Wearable Solid-State LiDAR Sensor
Publication authorsH. Liu, R. Liu, K. Yang, J. Zhang, K. Peng, R. Stiefelhagen
Publication URLhttps://arxiv.org/abs/2107.03180
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)Python with Cuda
HardwareRTX 3090
Submission creation date5 Jul, 2021
Last edited8 Jul, 2021

3D semantic instance results



Infoavg ap 50%bathtubbedbookshelfcabinetchaircountercurtaindeskdoorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwindow
0.6350.6670.7970.7140.5620.7740.1460.8100.4290.4760.5460.3990.6331.0000.6320.7220.6091.0000.514