SqueezeSegV3: Spatially-Adaptive Convolution for Efficient Point-Cloud Segmentation

  title={SqueezeSegV3: Spatially-Adaptive Convolution for Efficient Point-Cloud Segmentation},
  author={Chenfeng Xu and B. Wu and Z. Wang and Wei Zhan and P. Vajda and K. Keutzer and M. Tomizuka},
  • Chenfeng Xu, B. Wu, +4 authors M. Tomizuka
  • Published in ECCV 2020
  • Computer Science
  • LiDAR point-cloud segmentation is an important problem for many applications. For large-scale point cloud segmentation, the \textit{de facto} method is to project a 3D point cloud to get a 2D LiDAR image and use convolutions to process it. Despite the similarity between regular RGB and LiDAR images, we discover that the feature distribution of LiDAR images changes drastically at different image locations. Using standard convolutions to process such LiDAR images is problematic, as convolution… CONTINUE READING
    19 Citations

    Figures, Tables, and Topics from this paper

    Explore Further: Topics Discussed in This Paper

    Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation
    • PDF
    Cylinder3D: An Effective 3D Framework for Driving-scene LiDAR Semantic Segmentation
    • 3
    • PDF
    ePointDA: An End-to-End Simulation-to-Real Domain Adaptation Framework for LiDAR Point Cloud Segmentation
    • 1
    • PDF
    SalsaNext: Fast, Uncertainty-Aware Semantic Segmentation of LiDAR Point Clouds
    • 13
    • PDF
    Cascaded Non-local Neural Network for Point Cloud Semantic Segmentation
    • 1
    • PDF


    SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud
    • 127
    • PDF
    PointCNN: Convolution On X-Transformed Points
    • 494
    • PDF
    RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds
    • Q. Hu, B. Yang, +5 authors Andrew Markham
    • Computer Science, Engineering
    • 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
    • 2020
    • 85
    • PDF
    VV-Net: Voxel VAE Net With Group Convolutions for Point Cloud Segmentation
    • 30
    • PDF
    SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences
    • 135
    • Highly Influential
    • PDF
    End-to-End Multi-View Fusion for 3D Object Detection in LiDAR Point Clouds
    • 41
    • PDF
    RangeNet ++: Fast and Accurate LiDAR Semantic Segmentation
    • 76
    • Highly Influential
    • PDF
    SqueezeSeg: Convolutional Neural Nets with Recurrent CRF for Real-Time Road-Object Segmentation from 3D LiDAR Point Cloud
    • 239
    • PDF
    Learning to Sample
    • 26
    • PDF