AmodalSynthDrive: A Synthetic Amodal Perception Dataset for Autonomous Driving

TL;DR

AmodalSynthDrive is a comprehensive synthetic dataset designed to advance amodal perception in autonomous driving by providing multi-view images, 3D annotations, and diverse scenarios, enabling improved scene understanding and benchmarking.

Contribution

The paper introduces AmodalSynthDrive, a large-scale synthetic dataset for amodal perception in autonomous driving, addressing data scarcity and annotation challenges in occluded object understanding.

Findings

Baseline evaluations reveal significant challenges in amodal perception tasks.

The dataset enables benchmarking across multiple perception tasks.

Diverse conditions improve robustness of perception algorithms.

Abstract

Unlike humans, who can effortlessly estimate the entirety of objects even when partially occluded, modern computer vision algorithms still find this aspect extremely challenging. Leveraging this amodal perception for autonomous driving remains largely untapped due to the lack of suitable datasets. The curation of these datasets is primarily hindered by significant annotation costs and mitigating annotator subjectivity in accurately labeling occluded regions. To address these limitations, we introduce AmodalSynthDrive, a synthetic multi-task multi-modal amodal perception dataset. The dataset provides multi-view camera images, 3D bounding boxes, LiDAR data, and odometry for 150 driving sequences with over 1M object annotations in diverse traffic, weather, and lighting conditions. AmodalSynthDrive supports multiple amodal scene understanding tasks including the introduced amodal depth estimation for enhanced spatial understanding. We evaluate several baselines for each of these tasks to illustrate the challenges and set up public benchmarking servers. The dataset is available at http://amodalsynthdrive.cs.uni-freiburg.de.