KITTI

Dataset Information
Modalities
Images, LiDAR
Languages
English, Chinese
Introduced
2012
License
Homepage

Overview

KITTI (Karlsruhe Institute of Technology and Toyota Technological Institute) is one of the most popular datasets for use in mobile robotics and autonomous driving. It consists of hours of traffic scenarios recorded with a variety of sensor modalities, including high-resolution RGB, grayscale stereo cameras, and a 3D laser scanner. Despite its popularity, the dataset itself does not contain ground truth for semantic segmentation. However, various researchers have manually annotated parts of the dataset to fit their necessities. Álvarez et al. generated ground truth for 323 images from the road detection challenge with three classes: road, vertical, and sky. Zhang et al. annotated 252 (140 for training and 112 for testing) acquisitions – RGB and Velodyne scans – from the tracking challenge for ten object categories: building, sky, road, vegetation, sidewalk, car, pedestrian, cyclist, sign/pole, and fence. Ros et al. labeled 170 training images and 46 testing images (from the visual odometry challenge) with 11 classes: building, tree, sky, car, sign, road, pedestrian, fence, pole, sidewalk, and bicyclist.

Source: A Review on Deep Learning Techniques Applied to Semantic Segmentation
Image Source: http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d

Variants: KITTI Test (Offline Methods), KITTI Test (Online Methods), KITTI Cyclists Moderate val, KITTI (trained on 3DMatch), KITTI 2015 (train), KITTI2015 - 4x upscaling, KITTI2015 - 2x upscaling, KITTI2012 - 4x upscaling, KITTI2012 - 2x upscaling, KITTI2012 - 2x upscaling, KITTI2012 - 2x scaling, KITTI Pedestrian Moderate, KITTI Eigen Split Improved Ground Truth, KITTI Cyclist Moderate, KITTI Object Tracking Evaluation 2012, KITTI Pedestrian Easy, KITTI 2015 (train) , 2D KITTI Pedestrians Moderate, 2D KITTI Pedestrians Hard, 2D KITTI Pedestrians Easy, 2D KITTI Cyclists Moderate, 2D KITTI Cyclists Hard, 2D KITTI Cyclists Easy, Kitti Odometry, KITTI Cyclist Hard, KITTI Cyclist Easy, KITTI2012 Tracking, 2D KITTI Cars Hard, 2D KITTI Cars Easy, KITTI Stereo 2015, KITTI Stereo 2012, Kitti Raw, KITTI Pedestrian, KITTI Cars Simple, KITTI (FCGF setting), 2D KITTI Cars Moderate, KITTI 2015 Scene Flow Test, KITTI 2015 Scene Flow Training, KITTI Novel View Synthesis, KITTI 2015 - unsupervised, KITTI 2012 - unsupervised, KITTI Pedestrians Moderate val, KITTI Pedestrian Hard, KITTI Panoptic Segmentation, KITTI Horizon, KITTI Tracking test, KITTI2015, KITTI 2015 unsupervised, KITTI 2015 - 4x upscaling, KITTI 2015 - 2x upscaling, KITTI 2015, KITTI2012, KITTI 2012 unsupervised, KITTI 2012 - 4x upscaling, KITTI 2012 - 2x upscaling, KITTI 2012, KITTI Semantic Segmentation, KITTI Pedestrians Moderate, KITTI Pedestrians Hard, KITTI Pedestrians Easy, KITTI Pedestrian Moderate val, KITTI Pedestrian Hard val, KITTI Pedestrian Easy val, KITTI Eigen split unsupervised, KITTI Eigen split, KITTI Cyclists Moderate, KITTI Cyclists Hard, KITTI Cyclists Easy, KITTI Cyclist Moderate val, KITTI Cyclist Hard val, KITTI Cyclist Easy val, KITTI Cars Moderate val, KITTI Cars Moderate, KITTI Cars Hard val, KITTI Cars Hard, KITTI Cars Easy val, KITTI Cars Easy, KITTI

Associated Benchmarks

This dataset is used in 11 benchmarks:

Recent Benchmark Submissions

Task Model Paper Date
Unsupervised Panoptic Segmentation CUPS (27 pseudo-classes) Scene-Centric Unsupervised Panoptic Segmentation 2025-04-02
Unsupervised Panoptic Segmentation CUPS (40 pseudo-classes) Scene-Centric Unsupervised Panoptic Segmentation 2025-04-02
Unsupervised Panoptic Segmentation CUPS (54 pseudo-classes) Scene-Centric Unsupervised Panoptic Segmentation 2025-04-02
Image Clustering TURTLE (CLIP + DINOv2) Let Go of Your Labels … 2024-06-11
Scene Generation GaussianCity GaussianCity: Generative Gaussian Splatting for … 2024-06-10
Knowledge Distillation TIE-KD (T: Adabins S: MobileNetV2) TIE-KD: Teacher-Independent and Explainable Knowledge … 2024-02-22
Unsupervised Panoptic Segmentation U2Seg Unsupervised Universal Image Segmentation 2023-12-28
Video Prediction DMVFN A Dynamic Multi-Scale Voxel Flow … 2023-03-17
Novel View Synthesis READ READ: Large-Scale Neural Scene Rendering … 2022-05-11
Object Tracking M2-Track Beyond 3D Siamese Tracking: A … 2022-03-03
Depth Completion FusionDepth Advancing Self-supervised Monocular Depth Learning … 2021-09-20
Object Tracking BAT Box-Aware Feature Enhancement for Single … 2021-08-10
Visual Place Recognition None SSC: Semantic Scan Context for … 2021-07-01
Point Cloud Registration GeDi Learning general and distinctive 3D … 2021-05-21
Point Cloud Registration SpinNet SpinNet: Learning a General Surface … 2020-11-24
Point Cloud Registration DIP Distinctive 3D local deep descriptors 2020-09-01
Image Dehazing LCA LCA-Net: Light Convolutional Autoencoder for … 2020-08-24
Video Prediction FVS Future Video Synthesis with Object … 2020-04-01
Point Cloud Registration D3Feat-pred D3Feat: Joint Learning of Dense … 2020-03-06
Image Dehazing FFA-Net FFA-Net: Feature Fusion Attention Network … 2019-11-18

Research Papers

Recent papers with results on this dataset: