Kitti

KITTI is a public dataset and benchmark for autonomous driving research, named after the Karlsruhe Institute of Technology (KIT) and the Toyota Technological Institute at Chicago (TTI-Chicago). Introduced in the early 2010s, it has become a standard resource for evaluating computer vision and robotics algorithms in real-world driving scenarios.

The dataset is collected from a vehicle equipped with a stereo camera pair, a 64-beam LiDAR sensor, and GPS/IMU. It captures a variety of driving environments, including urban, rural, and highway scenes around Karlsruhe, Germany, providing synchronized, calibrated sensor data alongside precise ground-truth information.

KITTI supports multiple perception and localization tasks. Annotations and benchmarks cover stereo matching, optical flow, visual odometry, 3D object detection, and 3D tracking, with ground-truth data intended for objective evaluation. A separate semantic segmentation benchmark has also been developed as part of the suite.

Data are released in training and test splits to enable fair, reproducible comparisons among approaches. Over the years, KITTI has become widely used to train, validate, and compare perception and SLAM systems, influencing the direction of research in 3D perception, autonomous driving, and benchmarking practices.