A Background Estimation dataset
Presentation
These pages describe a Background Estimation dataset (BEds), a corpus of video sequences generated from publicly available video-surveillance datasets to cover several Background Estimation challenges.
The dataset is focused on 4 challenges or categories conformed by 10 video sequences each and the associated ground-truth background image.
Video sequences have been extracted from the following public datasets related with the video-surveillance task:
1) MALL: Video sequences used for crowd counting and profiling research.
2) Bank Street: Video sequences used for anomaly detection.
3) CAVIAR: Video sequences used for event detection.
4) CUHK Crowd Dataset: Video sequences used for Scene-Independent Group Profiling in Crowd.
5) I2R: Video sequences used for Background Subtraction.
6) ATON: Video sequences used for Shadow detection.
7) CVRL: Video sequences used for video-surveillance in indoor scenarios.
8) IITK: Video sequences used for object-tracking, activity recognition, congestion analysis or vehicle counting.
9) CDNET2014: Video sequences used for Background Subtraction.
10) APIDIS: Video sequences used for detection and tracking.
11) AVSS2007: Video sequences used for abandoned object detection an illegally parked vehicle detection.
12) PBI: Video sequences recorded for Background Estimation.
13) Wallflower: Video sequences used for Background Subtraction
14) PETS2009: Video sequences used for pedestrian detection and tracking.
15) LOST: Video sequences recorded for long term observation of scenarios.
16) LIMU: Video sequences used for Background Subtraction.
For more details please refer to the "Content" section.
Related works (using dataset) :
[1]
Carolina Fernández-Pedraza, “Reconstrucción de fondo de escena a partir de secuencias de vídeo&rdquo, Graduate thesis of Grado en IngenierÃa de Tecnologías
y Servicios de Telecomunicación at Escuela Politécnica Superior (Universidad Autónoma de Madrid). Supervised by Diego Ortego and José M. Martínez.
Work partially supported by the Spanish Goverment under project TEC2014-53176-R (HAVideo).
