Demo: Real-time contour-based pedestrian detection

Real-time pedestrian detection in crowded scenarios still represents a major scientific challenge. Dynamic occlusions between humans and the presence of dense gradient structure (clutter) typically render such scenarios complex for automated visual analysis. In this demo we present an algorithmic framework which efficiently computes pedestrian-specific shape and motion cues and combines them in a probabilistic manner to infer the location and occlusion status of pedestrians viewed by a stationary camera. The articulated pedestrian shape is represented by a sparse contour template, where fast template matching against image features is carried out using integral images built along oriented scan-lines. The motion cue is computed by employing a non-parametric background model using the YCbCr color space. Given the probabilistic output from the two cues the spatial configuration of hypothesized human body locations is obtained by an iterative optimization scheme taking into account the depth ordering and occlusion status of individual hypotheses. The method achieves fast computation times even in complex scenarios with a high pedestrian density. The underlying algorithms have been heavily optimized. Furthermore, if GPGPU hardware is available, computationally expensive algorithmic parts are carried out on the GPU. This demo will demonstrate human detection on a set of complex scenes along with specific detection performance and run-time benchmarks.