M3D-RPN: Monocular 3D Region Proposal Network for Object Detection (2019-07-13T00:00:00.000000Z)

TL;DR

M3D-RPN is able to significantly improve the performance of both monocular 3D Object Detection and Bird's Eye View tasks within the KITTI urban autonomous driving dataset, while efficiently using a shared multi-class model.

Abstract

Understanding the world in 3D is a critical component of urban autonomous driving. Generally, the combination of expensive LiDAR sensors and stereo RGB imaging has been paramount for successful 3D object detection algorithms, whereas monocular image-only methods experience drastically reduced performance. We propose to reduce the gap by reformulating the monocular 3D detection problem as a standalone 3D region proposal network. We leverage the geometric relationship of 2D and 3D perspectives, allowing 3D boxes to utilize well-known and powerful convolutional features generated in the image-space. To help address the strenuous 3D parameter estimations, we further design depth-aware convolutional layers which enable location specific feature development and in consequence improved 3D scene understanding. Compared to prior work in monocular 3D detection, our method consists of only the proposed 3D region proposal network rather than relying on external networks, data, or multiple stages. M3D-RPN is able to significantly improve the performance of both monocular 3D Object Detection and Bird's Eye View tasks within the KITTI urban autonomous driving dataset, while efficiently using a shared multi-class model.

Authors

Xiaoming Liu

9 papers

Garrick Brazil

4 papers

TL;DR

Abstract

Authors

References45 items

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

Pedestrian Detection With Autoregressive Network Phases

Deep Continuous Fusion for Multi-sensor 3D Object Detection

Context Refinement for Object Detection

Bi-box Regression for Pedestrian Detection and Occlusion Estimation

DetNet: Design Backbone for Object Detection

Parallel Feature Pyramid Network for Object Detection

Learning Efficient Single-Stage Pedestrian Detectors by Asymptotic Localization Fitting

Deep Feature Pyramid Reconfiguration for Object Detection

Localization Recall Precision (LRP): A New Performance Metric for Object Detection

Multi-level Fusion Based 3D Object Detection from Monocular Images

SurfConv: Bridging 3D and 2D Convolution for RGBD Images

PIXOR: Real-time 3D Object Detection from Point Clouds

LiDAR-Video Driving Dataset: Learning Driving Policies Effectively

Tiny SSD: A Tiny Single-Shot Detection Deep Convolutional Neural Network for Real-Time Embedded Object Detection

A General Pipeline for 3D Detection of Vehicles

Cascade R-CNN: Delving Into High Quality Object Detection

Single-Shot Object Detection with Enriched Semantics

Frustum PointNets for 3D Object Detection from RGB-D Data

Repulsion Loss: Detecting Pedestrians in a Crowd

Dynamic Zoom-in Network for Fast Object Detection in Large Images

Automatic differentiation in PyTorch

SnapNet-R: Consistent 3D Multi-view Semantic Labeling for Robotics

Bounding Boxes, Segmentations and Object Coordinates: How Important is Recognition for 3D Scene Flow Estimation in Autonomous Driving Scenarios?

Monocular Video-Based Trailer Coupler Detection Using Multiplexer Convolutional Neural Network

Illuminating Pedestrians via Simultaneous Detection and Segmentation

Accurate Single Stage Detector Using Recurrent Rolling Convolution

CNN-SLAM: Real-Time Dense Monocular SLAM with Learned Depth Prediction

3D Bounding Box Estimation Using Deep Learning and Geometry

Multi-view 3D Object Detection Network for Autonomous Driving

Densely Connected Convolutional Networks

UnitBox: An Advanced Object Detection Network

A Unified Multi-scale Deep Convolutional Neural Network for Fast Object Detection

Monocular 3D Object Detection for Autonomous Driving

Exploit All the Layers: Fast and Accurate CNN Object Detector with Scale Dependent Pooling and Cascaded Rejection Classifiers

Subcategory-Aware Convolutional Neural Networks for Object Proposals and Detection

SSD: Single Shot MultiBox Detector

3D Object Proposals for Accurate Object Class Detection

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

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

Fast R-CNN

ImageNet classification with deep convolutional neural networks

Are we ready for autonomous driving? The KITTI vision benchmark suite

Gradient Feature Selection for Online Boosting

Object Recognition with Gradient-Based Learning

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names