Fast Point R-CNN

By Yilun Chen, Yilun Chen, Shu Liu, Xiaoyong Shen, Jiaya Jia. 2019

445 citations recorded by the source corpus.

We present a unified, efficient and effective framework for point-cloud based 3D object detection. Our two-stage approach utilizes both voxel representation and raw point cloud data to exploit respective advantages. The first stage network, with voxel representation as input, only consists of light convolutional operations, producing a small number of high-quality initial predictions. Coordinate and indexed convolutional feature of each point in initial prediction are effectively fused with the attention mechanism, preserving both accurate localization and context information. The second stage works on interior points with their fused feature for further refining the prediction. Our method is evaluated on KITTI dataset, in terms of both 3D and Bird's Eye View (BEV) detection, and achieves state-of-the-arts with a 15FPS detection rate.

Subfields: Computational Mechanics, Environmental Engineering, Geology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Environmental Engineering

In Geology

Citations in the corpus

3D-CVF: Generating Joint Camera and LiDAR Features Using Cross-View Spatial Feature Fusion for 3D Object Detection (2020, background): “32 Fast PointRCNN [3] LiDAR 65 85.”
RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds (2019, background): “Alternatively, point clouds can be voxelized into 3D grids and then powerful 3D CNNs are applied in [14, 28, 10, 39, 9].”
3DSSD: Point-Based 3D Single Stage Object Detector (2020, unknown): “3] 82.19 69.79 60.59 AVOD-FPN [12] 83.07 71.76 65.73 IPOD [34] 80.30 73.04 68.73 UberATG-MMF [16] 88.40 77.43 70.22 F-ConvNet [30] 87.36 76.39 66.69 PointRCNN [26] L 86.96 75.64 70.70 Fast Point-RCNN [5] 85.29 77.40 70.24 Patches [14] 88.67 77.20 71.82 MMLab-PartAˆ2 [27] 87.81 78.49 73.51 STD [35] 8”
Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud (2020, methodology): “For the ablation study, we follow the standard practice [10][21][5] and split the training samples into a training split of 3712 samples and a validation split of 3769 samples.”
Learning Depth-Guided Convolutions for Monocular 3D Object Detection (2019, background): “[44, 7] directly generated a small number of high-quality 3D proposals from point clouds via segmenting the point clouds of the whole scene into foreground and background.”
EPNet: Enhancing Point Features with Image Semantics for 3D Object Detection (2020, unknown): “53 Fast Point R-CNN [4] L 85.”
SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloud (2021, background): “26 80* Fast PointRCNN [4] ICCV 2019 LiDAR 85.”
Center-based 3D Object Detection and Tracking (2020, methodology): “Recent works considered directly applying RCNN style 2D detectors to the 3D domains [9,44–46 , 61 ].”
Structure Aware Single-Stage 3D Object Detection From Point Cloud (2020, methodology): “2, we employ the commonly used backbone network [30, 25, 2] as our feature extractor.”
Not All Points Are Equal: Learning Highly Efficient Point-based Detectors for 3D LiDAR Point Clouds (2022, unknown): “Po in tVo xe l Fast Point R-CNN [3] ICCV 2019 2-stage 79.”
BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation (2022, background): “Another stream of research [24], [25], [26], [27], [28], [29] focuses on twostage object detector design, which adds an RCNN network to existing one-stage object detectors.”
3D Object Detection for Autonomous Driving: A Comprehensive Survey (2022, background): “Fast Point R-CNN [44] RefinerNet PV-RCNN [253] RoI-grid Pooling PV-RCNN++ [255] VectorPool Pyramid R-CNN [185] RoI-grid Attention LiDAR R-CNN [144] Scale-aware Pooling CT3D [250] Channel-wise Transformer”
PV-RCNN++: Point-Voxel Feature Set Abstraction With Local Vector Representation for 3D Object Detection (2021, methodology): “In addition, (Wang et al., 2020; Chen et al., 2019a) predicts bounding box parameters following anchor-free paradigm.”
Towards Robust LiDAR-based Perception in Autonomous Driving: General Black-box Adversarial Sensor Attack and Countermeasures (2020, background): “Instead of transforming point clouds to regular 2D structures or voxels for feature extraction, recent studies propose to directly operate on point clouds for 3D object detection [25, 53, 54, 71] and achieve the state-of-the-art performance.”