Introduction

KITTIOne of the most important test sets in the field of automated driving at present, KITTI focuses on image processing technologies for automated driving, mainly applied to automated driving perception and prediction, which also involves localization and SLAM technologies.

We develop novel and challenging real-world computer vision benchmarks using our autonomous driving platform Annieway. The tasks we are interested in are: stereo, optical flow, visual odometry, 3D object detection and 3D tracking. For this purpose, we equip a standard station wagon with two high-resolution color and grayscale cameras. a Velodyne laser scanner and a GPS positioning system provide accurate ground truth. Our dataset was captured by driving around the medium-sized city of Karlsruhe, in rural areas and on the highway. Up to 15 cars and 30 pedestrians are visible in each image. In addition to making all data available in raw format, we also extracted benchmarks for each task. For each of our benchmarks, we also provide an evaluation metric and this evaluation website. Preliminary experiments have shown that methods that rank high on established benchmarks such as Middlebury perform below average when transferred from the lab to the real world. Our goal is to reduce this bias and complement existing benchmarks by providing the community with real-world benchmarks with new difficulties.

Registration Process

Here you can:(1) Register to download KITTIas well as(2) Application submitted to KITTI. The second step is optional and usually takes place at a later point in time. The second step can only be performed if the paper is submitted to a conference where the experimental results have been prepared and the conference has been finalized. Submissions will be reviewed manually and may take up to one week.

To register to download the KITTI dataset, enter all information below, including your institutional email address (e.g. *.edu). You will receive an email within a few minutes with a confirmation link to activate your account. Check your spam folder. Please note that you cannot apply for multiple accounts with multiple email addresses.You can only register with one e-mail address! Otherwise we will ban you.

We open the KITTI dataset registration site at http://www.cvlibs.net/datasets/kitti/user_login.php如下图所示: we click on the bottom of theRegister here

Successful registration

Our edu email will receive an Account Activation account activation email from KITTI Evaluation Benchmark.

Warning: This email originated outside Santa Fe College. Consider carefully before sharing information.

Please use this link to verify your identity and activate your account.
http://www.cvlibs.net/datasets/kitti/user_activate.php?user=6538f620d33384475111111
Once activated, you can download the KITTI datasets. Note: Please do not complete registration step 2 unless you want to publish a paper, have developed and implemented your method and want to submit your results for a conference/journal in the very near future. Please do not complete registration step 2 unless you want to publish a paper, have developed and implemented your method and want to submit your results for a conference/journal in the very near future. If you only want to download the KITTI dataset for training or evaluation, you are done here. You can now login and download the dataset.

After activation, you can download the KITT I dataset. Note: Please do not complete registration step 2 unless you want to publish a paper, have developed and implemented your method, and want to submit your results for a conference/journal in the near future. If you just want to download the KITT I dataset for training or evaluation, you can do so here. You can now login and download the dataset.