Multi-View Representation is What You Need for Point-Cloud Pre-Training

• URL: http://arxiv.org/abs/2306.02558v3
• Date: Sun, 28 Apr 2024 18:44:33 GMT
• Title: Multi-View Representation is What You Need for Point-Cloud Pre-Training
• Authors: Siming Yan, Chen Song, Youkang Kong, Qixing Huang,
• Abstract summary: This paper proposes a novel approach to point-cloud pre-training that learns 3D representations by leveraging pre-trained 2D networks. We train the 3D feature extraction network with the help of the novel 2D knowledge transfer loss. Experimental results demonstrate that our pre-trained model can be successfully transferred to various downstream tasks.
• Score: 22.55455166875263
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: A promising direction for pre-training 3D point clouds is to leverage the massive amount of data in 2D, whereas the domain gap between 2D and 3D creates a fundamental challenge. This paper proposes a novel approach to point-cloud pre-training that learns 3D representations by leveraging pre-trained 2D networks. Different from the popular practice of predicting 2D features first and then obtaining 3D features through dimensionality lifting, our approach directly uses a 3D network for feature extraction. We train the 3D feature extraction network with the help of the novel 2D knowledge transfer loss, which enforces the 2D projections of the 3D feature to be consistent with the output of pre-trained 2D networks. To prevent the feature from discarding 3D signals, we introduce the multi-view consistency loss that additionally encourages the projected 2D feature representations to capture pixel-wise correspondences across different views. Such correspondences induce 3D geometry and effectively retain 3D features in the projected 2D features. Experimental results demonstrate that our pre-trained model can be successfully transferred to various downstream tasks, including 3D shape classification, part segmentation, 3D object detection, and semantic segmentation, achieving state-of-the-art performance.

Related papers

• ConDense: Consistent 2D/3D Pre-training for Dense and Sparse Features from Multi-View Images [47.682942867405224]
  ConDense is a framework for 3D pre-training utilizing existing 2D networks and large-scale multi-view datasets. We propose a novel 2D-3D joint training scheme to extract co-embedded 2D and 3D features in an end-to-end pipeline.
  arXiv  Detail & Related papers  (2024-08-30T05:57:01Z)
• 3DiffTection: 3D Object Detection with Geometry-Aware Diffusion Features [70.50665869806188]
  3DiffTection is a state-of-the-art method for 3D object detection from single images. We fine-tune a diffusion model to perform novel view synthesis conditioned on a single image. We further train the model on target data with detection supervision.
  arXiv  Detail & Related papers  (2023-11-07T23:46:41Z)
• SSR-2D: Semantic 3D Scene Reconstruction from 2D Images [54.46126685716471]
  In this work, we explore a central 3D scene modeling task, namely, semantic scene reconstruction without using any 3D annotations. The key idea of our approach is to design a trainable model that employs both incomplete 3D reconstructions and their corresponding source RGB-D images. Our method achieves the state-of-the-art performance of semantic scene completion on two large-scale benchmark datasets MatterPort3D and ScanNet.
  arXiv  Detail & Related papers  (2023-02-07T17:47:52Z)
• MvDeCor: Multi-view Dense Correspondence Learning for Fine-grained 3D Segmentation [91.6658845016214]
  We propose to utilize self-supervised techniques in the 2D domain for fine-grained 3D shape segmentation tasks. We render a 3D shape from multiple views, and set up a dense correspondence learning task within the contrastive learning framework. As a result, the learned 2D representations are view-invariant and geometrically consistent.
  arXiv  Detail & Related papers  (2022-08-18T00:48:15Z)
• Multi-Modality Task Cascade for 3D Object Detection [22.131228757850373]
  Many methods train two models in isolation and use simple feature concatenation to represent 3D sensor data. We propose a novel Multi-Modality Task Cascade network (MTC-RCNN) that leverages 3D box proposals to improve 2D segmentation predictions. We show that including a 2D network between two stages of 3D modules significantly improves both 2D and 3D task performance.
  arXiv  Detail & Related papers  (2021-07-08T17:55:01Z)
• Learning from 2D: Pixel-to-Point Knowledge Transfer for 3D Pretraining [21.878815180924832]
  We present a novel 3D pretraining method by leveraging 2D networks learned from rich 2D datasets. Our experiments show that the 3D models pretrained with 2D knowledge boost the performances across various real-world 3D downstream tasks.
  arXiv  Detail & Related papers  (2021-04-10T05:40:42Z)
• 3D-to-2D Distillation for Indoor Scene Parsing [78.36781565047656]
  We present a new approach that enables us to leverage 3D features extracted from large-scale 3D data repository to enhance 2D features extracted from RGB images. First, we distill 3D knowledge from a pretrained 3D network to supervise a 2D network to learn simulated 3D features from 2D features during the training. Second, we design a two-stage dimension normalization scheme to calibrate the 2D and 3D features for better integration. Third, we design a semantic-aware adversarial training model to extend our framework for training with unpaired 3D data.
  arXiv  Detail & Related papers  (2021-04-06T02:22:24Z)
• Cylinder3D: An Effective 3D Framework for Driving-scene LiDAR Semantic Segmentation [87.54570024320354]
  State-of-the-art methods for large-scale driving-scene LiDAR semantic segmentation often project and process the point clouds in the 2D space. A straightforward solution to tackle the issue of 3D-to-2D projection is to keep the 3D representation and process the points in the 3D space. We develop a 3D cylinder partition and a 3D cylinder convolution based framework, termed as Cylinder3D, which exploits the 3D topology relations and structures of driving-scene point clouds.
  arXiv  Detail & Related papers  (2020-08-04T13:56:19Z)
• Self-supervised Feature Learning by Cross-modality and Cross-view Correspondences [32.01548991331616]
  This paper presents a novel self-supervised learning approach to learn both 2D image features and 3D point cloud features. It exploits cross-modality and cross-view correspondences without using any annotated human labels. The effectiveness of the learned 2D and 3D features is evaluated by transferring them on five different tasks.
  arXiv  Detail & Related papers  (2020-04-13T02:57:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.