KITTI-C
收藏资源简介:
KITTI-C数据集是基于KITTI数据集构建的,用于评估AI驱动的多传感器融合系统的性能和鲁棒性。该数据集包含多种感知任务,如目标检测、目标跟踪和深度完成,并设计了14种常见的真实损坏模式来合成大规模损坏数据集。KITTI-C旨在解决自动驾驶等安全关键应用中感知系统的性能和可靠性问题,通过系统评估帮助理解和改进多传感器融合系统的鲁棒性。
The KITTI-C dataset is constructed upon the KITTI dataset, and is utilized to evaluate the performance and robustness of AI-powered multi-sensor fusion systems. This dataset encompasses multiple perception tasks including object detection, object tracking and depth completion, with 14 common real-world corruption patterns employed to synthesize large-scale corrupted datasets. KITTI-C aims to address the performance and reliability challenges of perception systems in safety-critical applications such as autonomous driving, and facilitates the understanding and improvement of the robustness of multi-sensor fusion systems through systematic evaluation.
- 1Benchmarking Robustness of AI-Enabled Multi-sensor Fusion Systems: Challenges and Opportunities南京大学新型软件技术国家重点实验室 · 2023年



