Lab Database

Description

This database contains six sequences and their corresponding ground truth, which allows to use them to evaluate the quality of the results provided by detection and tracking strategies.

These sequences have been recorded in an indoor environment (specifically in a laboratory) and contain critical situations: dynamic backgrounds, shadows, illumination changes, moving objects remaining static different periods of time, and multiple moving objects similar to background regions.

A set of hand-segmented evaluation images (ground truth images) can be also obtained in conjuntion with each of the sequences in this database. These hand-segmented images are uniformly distributed along the sequences (one every 25 frames) and make it possible to obtain qualitative data to evaluate the performance of different detection and tracking strategies.

A .rar file for each of the six test sequences can be downloaded. Each file contains two folders named respectively "Original images" and "Ground truth". The first one contains one 24 bpp BMP file for each frame in the original video. The second one contains 1 bpp BMP files with the ground truth corresponding to some images of the sequence.

For questions about these test images, please contact Carlos Cuevas at This email address is being protected from spambots. You need JavaScript enabled to view it..

Citation

C. Cuevas, N. García, “Efficient moving object detection for lightweight applications on smart cameras”, IEEE Trans. Circuits and Systems for Video Technology, vol. 23, no. 1, pp. 1-14, 2013.

Sequences

Lab_001

One moving object crosses the laboratory causing many shadows and reflections. Moreover, the moving object contains regions very similar to some background areas that it crosses.



Download

Lab_002

Two moving objects cross the scene. In half of their trayectories they interact with one another, and remain stopped for a few seconds before continuing their ways.