Real-time Pedestrian Surveillance with Top View Cumulative Grids
This work addresses real-time surveillance for security applications, but it is incremental as it builds on existing scene flow methods with new sensor data.
The paper tackles the problem of real-time pedestrian surveillance by mapping footage to an aerial view using cumulative grids on top view scene flow, tested on multiview RGB and RGB-D sensor footage.
This manuscript presents an efficient approach to map pedestrian surveillance footage to an aerial view for global assessment of features. The analysis of the footages relies on low level computer vision and enable real-time surveillance. While we neglect object tracking, we introduce cumulative grids on top view scene flow visualization to highlight situations of interest in the footage. Our approach is tested on multiview footage both from RGB cameras and, for the first time in the field, on RGB-D-sensors.