UNLV Pedestrian Dataset (UNLV-Peds):
UNLV Pedestrian dataset is specifically provided for video surveillance by leveraging available traffic cameras in the Las Vegas area.  The dataset has been prepared by Mohammad Shokrolah Shirazi, and Brendan Morris for the behavior and safety analysis of intersections. 

Dataset Characteristics:
- It has total 3000 pedestrian samples taken from 18 different intersections at Las Vegas.
- The data-set has 900 image files, no scale normalization with .bmp extension.
- Suppose the image files are located inside the /rawdata folder, the annotation file structure is as follows:
/rawdata/image_file.bmp #pedestrians x y w h x y w h ... 
where [x y w h] shows the bounding box coordinates for each pedestrian sample, x, y corresponds to top left coordinates, and w, h are width and height of the bounding box containing a pedestrian. 


Please cite the following paper if you use the data-set for your project/research:

M. S. Shirazi and B. T. Morris, "Vision-based pedestrian behavior analysis at intersections," Journal of Electronic Imaging, vol. 25, no. 5, pp. 1-13, March, 2016.

@article{shirazi_jei2016,
author = {Mohammad Shokrolah Shirazi and Brendan Tran Morris},
title = {{Vision-based pedestrian behavior analysis at intersections}},
volume = {25},
journal = {Journal of Electronic Imaging},
number = {5},
publisher = {SPIE},
pages = {1 -- 13},
keywords = {vision-based data collection system, pedestrian behavior, pedestrian tracking system, contextual fusion, intersection videos, Video surveillance, Video, Sensors, Optical flow, Cameras, Optical tracking, Lanthanum, Surveillance, Detection and tracking algorithms, Motion detection},
year = {2016},
doi = {10.1117/1.JEI.25.5.051203},
URL = {https://doi.org/10.1117/1.JEI.25.5.051203}
}