LPR and MMR Database Information

Prerequisites

All provided code is in C++ and makes use of the OpenCV library for storing images and various image operations. You will need to download and install it before being able to use the provided code.

Note: This database is password protected; if you would like to have access to it, please contact Louka Dlagnekov or Serge Belongie.

If you use this database in your research, please use the following citation format:

L. Dlagnekov and S. Belongie, ``UCSD/Calit2 Car License Plate, Make and Model Database,'' http://belongielab.org/car_data.html.

General Information

The provided data include:

1. About 10 hours of recorded video of cars entering the UCSD campus from the Gilman entrance during various times of day
2. Still frames taken from video feeds, hand-labeled with make and model information, license plate locations, and license plate texts
3. Frame by frame snapshots of the license plates of 878 cars as they are tracked from the moment they enter the camera's field of view to when they go out of range. These sequences are hand-labeled with license plate text and are stored in the same .dat format.
4. Still images of 291 cars in parking lots taken with a digital camera.

Download

The still frames may be downloaded from the following link:

stills.tar.gz    (183 MB)

Due to the large size of the raw videos, you may download these separately from the video subdirectory.

Raw Video

The file format of all video and .dat files is a sequence of JPEG images taken at 6 frames per second:

--myboundary\r
Content-Type: image/jpeg\r
\r
<JPEG IMAGE DATA>\r
--myboundary\r
Content-Type: image/jpeg\r
\r
<JPEG IMAGE DATA>\r
...

C++ code exists in the code/ subdirectory to read the video files frame by frame. The relevant files are VideoReader.h and VideoReader.cpp.

Most recently updated on Sep. 24, 2023 by Serge Belongie.