TransLoc3D : Point Cloud based Large-scale Place Recognition using Adaptive Receptive Fields (2021-05-25T00:00:00.000000Z)

TL;DR

A novel method named TransLoc3D is proposed, utilizing adaptive receptive fields with a point-wise reweighting scheme to handle objects of different sizes while suppressing noises, and an external transformer to capture long-range feature dependencies.

Abstract

Place recognition plays an essential role in the field of autonomous driving and robot navigation. Point cloud based methods mainly focus on extracting global descriptors from local features of point clouds. Despite having achieved promising results, existing solutions neglect the following aspects, which may cause performance degradation: (1) huge size difference between objects in outdoor scenes; (2) moving objects that are unrelated to place recognition; (3) long-range contextual information. We illustrate that the above aspects bring challenges to extracting discriminative global descriptors. To mitigate these problems, we propose a novel method named TransLoc3D, utilizing adaptive receptive fields with a point-wise reweighting scheme to handle objects of different sizes while suppressing noises, and an external transformer to capture long-range feature dependencies. As opposed to existing architectures which adopt fixed and limited receptive fields, our method benefits from size-adaptive receptive fields as well as global contextual information, and outperforms current state-of-the-arts with significant improvements on popular datasets.

Authors

Tianhan Xu

2 papers

Song-Hai Zhang

6 papers

Yu-Kun Lai

9 papers

TL;DR

Abstract

Authors

References71 items

Is Attention Better Than Matrix Decomposition?

Beyond Self-Attention: External Attention Using Two Linear Layers for Visual Tasks

MinkLoc++: Lidar and Monocular Image Fusion for Place Recognition

NDT-Transformer: Large-Scale 3D Point Cloud Localisation using the Normal Distribution Transform Representation

Transformer in Transformer

Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions

Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers

PCT: Point cloud transformer

End-to-End Video Instance Segmentation with Transformers

SOE-Net: A Self-Attention and Orientation Encoding Network for Point Cloud based Place Recognition

Rethinking Transformer-based Set Prediction for Object Detection

UP-DETR: Unsupervised Pre-training for Object Detection with Transformers

MinkLoc3D: Point Cloud Based Large-Scale Place Recognition

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

Generative Pretraining From Pixels

DAGC: Employing Dual Attention and Graph Convolution for Point Cloud based Place Recognition

End-to-End Object Detection with Transformers

A Metric Learning Reality Check

Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud

3DSSD: Point-Based 3D Single Stage Object Detector

Revisiting Training Strategies and Generalization Performance in Deep Metric Learning

RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds

ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks

PCAN: 3D Attention Map Learning Using Contextual Information for Point Cloud Based Retrieval

4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks

Relation-Shape Convolutional Neural Network for Point Cloud Analysis

Selective Kernel Networks

LPD-Net: 3D Point Cloud Learning for Large-Scale Place Recognition and Environment Analysis

PointRCNN: 3D Object Proposal Generation and Detection From Point Cloud

Scan Context: Egocentric Spatial Descriptor for Place Recognition Within 3D Point Cloud Map

Stochastic Attraction-Repulsion Embedding for Large Scale Image Localization

Texture Mapping for 3D Reconstruction with RGB-D Sensor

Delight: An Efficient Descriptor for Global Localisation Using LiDAR Intensities

PointNetVLAD: Deep Point Cloud Based Retrieval for Large-Scale Place Recognition

Dynamic Graph CNN for Learning on Point Clouds

3D Semantic Segmentation with Submanifold Sparse Convolutional Networks

Fine-Tuning CNN Image Retrieval with No Human Annotation

Efficient Global 2D-3D Matching for Camera Localization in a Large-Scale 3D Map

3D visual perception for self-driving cars using a multi-camera system: Calibration, mapping, localization, and obstacle detection

Sampling Matters in Deep Embedding Learning

Learnable pooling with Context Gating for video classification

Attention is All you Need

PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space

In Defense of the Triplet Loss for Person Re-Identification

1 year, 1000 km: The Oxford RobotCar dataset

Feature Pyramid Networks for Object Detection

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation

M2DP: A novel 3D point cloud descriptor and its application in loop closure detection

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

Robust and Efficient Relative Pose with a Multi-camera System for Autonomous Vehicle in Highly Dynamic Environments

Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning

Deep Residual Learning for Image Recognition

SSD: Single Shot MultiBox Detector

Rethinking the Inception Architecture for Computer Vision

NetVLAD: CNN Architecture for Weakly Supervised Place Recognition

You Only Look Once: Unified, Real-Time Object Detection

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

ORB-SLAM: A Versatile and Accurate Monocular SLAM System

Going deeper with convolutions

Very Deep Convolutional Networks for Large-Scale Image Recognition

3D ShapeNets: A deep representation for volumetric shapes

Motion Estimation for Self-Driving Cars with a Generalized Camera

Location Recognition Using Prioritized Feature Matching

Scan registration for autonomous mining vehicles using 3D‐NDT

Contrast's effect on spatial summation by macaque V1 neurons

Dynamic changes in receptive-field size in cat primary visual cortex.

Orientation-selective inhibition from beyond the classic visual receptive field

and Z

Appendix for : Graph Attention Convolution for Point Cloud Semantic Segmentation

Computer Vision Foundation / IEEE

Field of Study

Journal Information

Name

Page

Volume