Local Empirical Templates and Density Ratios for People Counting

Hung, Dao Huu; Chung, Sheng‐Luen; Hsu, Gee-Sern

doi:10.1007/978-3-642-19282-1_8

Cited by 4 publications

(15 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Interestingly, the obtained density ratio bounds given in Table 1 are quite similar to the ones in [18]. That is, it demonstrates the generality of density ratio bounds to cope with various viewpoints and a variety of degrees of human proximity.…”

Section: Offline Learning and 3d Human Models Based Simulationsupporting

confidence: 69%

“…Both lower and upper bounds of each category are so-called density ratio bounds. It seems that the bounds are invariant to camera viewpoints, lighting conditions, and the pedestrians' proximity [18]. The limited observed data, however undoubtedly, cannot cover all situations in reality to prove these above observations.…”

Section: Introductionmentioning

confidence: 94%

“…The training dataset used in the offline learning phase in this paper is completely different and more sufficient than the one used in [18] since it covers typical situations, in terms of camera viewpoints and the degree of human proximity. Interestingly, the obtained density ratio bounds given in Table 1 are quite similar to the ones in [18].…”

Section: Offline Learning and 3d Human Models Based Simulationmentioning

confidence: 99%

“…The number in the top-left corner of the image is the total estimated count throughout entire image. Manually counted ground truth of each frame is compared with the results of our proposed method, our previos work [18], holistic properties-based method [1] and moving corner-based method [2], tested on the same dataset. These results are adapted from their papers [1], [2].…”

Section: Evaluation On Pets 2009 Benchmarkmentioning

confidence: 99%

“…In the former approach reported in our previous work [18], occludedpedestrians patterns are extracted from a large collection of surveillance videos under different viewpoints and weather conditions. These patterns are manually classified into categories according to the number of people they contain, for computing the density ratios.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Real-Time Counting People in Crowded Areas by Using Local Empirical Templates and Density Ratios

Hung

Hsu

Chung

et al. 2012

IEICE Trans. Inf. & Syst.

Self Cite

View full text Add to dashboard Cite

SUMMARYIn this paper, a fast and automated method of counting pedestrians in crowded areas is proposed along with three contributions. We firstly propose Local Empirical Templates (LET), which are able to outline the foregrounds, typically made by single pedestrians in a scene. LET are extracted by clustering foregrounds of single pedestrians with similar features in silhouettes. This process is done automatically for unknown scenes. Secondly, comparing the size of group foreground made by a group of pedestrians to that of appropriate LET captured in the same image patch with the group foreground produces the density ratio. Because of the local scale normalization between sizes, the density ratio appears to have a bound closely related to the number of pedestrians who induce the group foreground. Finally, to extract the bounds of density ratios for groups of different number of pedestrians, we propose a 3D human models based simulation in which camera viewpoints and pedestrians' proximity are easily manipulated. We collect hundreds of typical occluded-people patterns with distinct degrees of human proximity and under a variety of camera viewpoints. Distributions of density ratios with respect to the number of pedestrians are built based on the computed density ratios of these patterns for extracting density ratio bounds. The simulation is performed in the offline learning phase to extract the bounds from the distributions, which are used to count pedestrians in online settings. We reveal that the bounds seem to be invariant to camera viewpoints and humans' proximity. The performance of our proposed method is evaluated with our collected videos and PETS 2009's datasets. For our collected videos with the resolution of 320x240, our method runs in real-time with good accuracy and frame rate of around 30 fps, and consumes a small amount of computing resources. For PETS 2009's datasets, our proposed method achieves competitive results with other methods tested on the same datasets [1], [2]. key words: local empirical templates, local density ratios, density ratio bounds, and people counting

show abstract

Section: Offline Learning and 3d Human Models Based Simulationsupporting

confidence: 69%

Section: Introductionmentioning

confidence: 94%