Cohere3D: Exploiting Temporal Coherence for Unsupervised Representation Learning of Vision-based Autonomous Driving

• URL: http://arxiv.org/abs/2402.15583v1
• Date: Fri, 23 Feb 2024 19:43:01 GMT
• Title: Cohere3D: Exploiting Temporal Coherence for Unsupervised Representation Learning of Vision-based Autonomous Driving
• Authors: Yichen Xie, Hongge Chen, Gregory P. Meyer, Yong Jae Lee, Eric M. Wolff, Masayoshi Tomizuka, Wei Zhan, Yuning Chai, Xin Huang
• Abstract summary: We propose a novel contrastive learning algorithm, Cohere3D, to learn coherent instance representations in a long-term input sequence. We evaluate our algorithm by finetuning the pretrained model on various downstream perception, prediction, and planning tasks.
• Score: 73.3702076688159
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Due to the lack of depth cues in images, multi-frame inputs are important for the success of vision-based perception, prediction, and planning in autonomous driving. Observations from different angles enable the recovery of 3D object states from 2D image inputs if we can identify the same instance in different input frames. However, the dynamic nature of autonomous driving scenes leads to significant changes in the appearance and shape of each instance captured by the camera at different time steps. To this end, we propose a novel contrastive learning algorithm, Cohere3D, to learn coherent instance representations in a long-term input sequence robust to the change in distance and perspective. The learned representation aids in instance-level correspondence across multiple input frames in downstream tasks. In the pretraining stage, the raw point clouds from LiDAR sensors are utilized to construct the long-term temporal correspondence for each instance, which serves as guidance for the extraction of instance-level representation from the vision-based bird's eye-view (BEV) feature map. Cohere3D encourages a consistent representation for the same instance at different frames but distinguishes between representations of different instances. We evaluate our algorithm by finetuning the pretrained model on various downstream perception, prediction, and planning tasks. Results show a notable improvement in both data efficiency and task performance.

Related papers

• VisionPAD: A Vision-Centric Pre-training Paradigm for Autonomous Driving [44.91443640710085]
  VisionPAD is a novel self-supervised pre-training paradigm for vision-centric algorithms in autonomous driving. It reconstructs multi-view representations using only images as supervision. It significantly improves performance in 3D object detection, occupancy prediction and map segmentation.
  arXiv  Detail & Related papers  (2024-11-22T03:59:41Z)
• IVT: An End-to-End Instance-guided Video Transformer for 3D Pose Estimation [6.270047084514142]
  Video 3D human pose estimation aims to localize the 3D coordinates of human joints from videos. IVT enables learningtemporal contextual depth information from visual features and 3D poses directly from video frames. Experiments on three widely-used 3D pose estimation benchmarks show that the proposed IVT achieves state-of-the-art performances.
  arXiv  Detail & Related papers  (2022-08-06T02:36:33Z)
• Neural Groundplans: Persistent Neural Scene Representations from a Single Image [90.04272671464238]
  We present a method to map 2D image observations of a scene to a persistent 3D scene representation. We propose conditional neural groundplans as persistent and memory-efficient scene representations.
  arXiv  Detail & Related papers  (2022-07-22T17:41:24Z)
• BEVerse: Unified Perception and Prediction in Birds-Eye-View for Vision-Centric Autonomous Driving [92.05963633802979]
  We present BEVerse, a unified framework for 3D perception and prediction based on multi-camera systems. We show that the multi-task BEVerse outperforms single-task methods on 3D object detection, semantic map construction, and motion prediction.
  arXiv  Detail & Related papers  (2022-05-19T17:55:35Z)
• Unsupervised View-Invariant Human Posture Representation [28.840986167408037]
  We present a novel unsupervised approach that learns to extract view-invariant 3D human pose representation from a 2D image. Our model is trained by exploiting the intrinsic view-invariant properties of human pose between simultaneous frames. We show improvements on the state-of-the-art unsupervised cross-view action classification accuracy on RGB and depth images.
  arXiv  Detail & Related papers  (2021-09-17T19:23:31Z)
• Spatio-temporal Self-Supervised Representation Learning for 3D Point Clouds [96.9027094562957]
  We introduce a-temporal representation learning framework, capable of learning from unlabeled tasks. Inspired by how infants learn from visual data in the wild, we explore rich cues derived from the 3D data. STRL takes two temporally-related frames from a 3D point cloud sequence as the input, transforms it with the spatial data augmentation, and learns the invariant representation self-supervisedly.
  arXiv  Detail & Related papers  (2021-09-01T04:17:11Z)
• 3D Neural Scene Representations for Visuomotor Control [78.79583457239836]
  We learn models for dynamic 3D scenes purely from 2D visual observations. A dynamics model, constructed over the learned representation space, enables visuomotor control for challenging manipulation tasks.
  arXiv  Detail & Related papers  (2021-07-08T17:49:37Z)
• Fine-Grained Vehicle Perception via 3D Part-Guided Visual Data Augmentation [77.60050239225086]
  We propose an effective training data generation process by fitting a 3D car model with dynamic parts to vehicles in real images. Our approach is fully automatic without any human interaction. We present a multi-task network for VUS parsing and a multi-stream network for VHI parsing.
  arXiv  Detail & Related papers  (2020-12-15T03:03:38Z)
• Self-Supervised Multi-View Synchronization Learning for 3D Pose Estimation [39.334995719523]
  Current methods cast monocular 3D human pose estimation as a learning problem by training neural networks on large data sets of images and corresponding skeleton poses. We propose an approach that can exploit small annotated data sets by fine-tuning networks pre-trained via self-supervised learning on (large) unlabeled data sets. We demonstrate the effectiveness of the synchronization task on the Human3.6M data set and achieve state-of-the-art results in 3D human pose estimation.
  arXiv  Detail & Related papers  (2020-10-13T08:01:24Z)

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.