Finding People in Images and Videos

By Navneet Dalal. 2006

387 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Efficient HOG human detection (2011, methodology): “The trilinear interpolation is given by [9] hðx1,y1,y1Þ’hðx1,y1,y1Þ þ9rf ðx,yÞ9 1 x x1 dx 1 y y1 dy 1 yðx,yÞ y1 dy , ð1Þ”
Pedestrian Detection: An Evaluation of the State of the Art (2012, background): “Two dominant nonmaximum suppression approaches have emerged: mean shift (MS) mode estimation [55] and pairwise max (PM) suppression [71], which discards the less confident of every pair of detections that overlap sufficiently according to (1).”
Exploring Weak Stabilization for Motion Feature Extraction (2013, background): “We show that one can exploit simple motion features to significantly increase detection accuracy with little additional computation.”
Survey of Pedestrian Detection for Advanced Driver Assistance Systems (2010, background): “Assuming that classifiers provide a peak at the correct position and scale of the target and weaker responses around it, Dalal [110] makes use of mean shift [111] to find the minimum set of ROIs that best adjust to the pedestrians in the image.”
An HOG-LBP human detector with partial occlusion handling (2009, unknown): “In order to take the advantage of the integral image without impairing the performance, we propose an approach, named as Convoluted Trilinear Interpolation (CTI), to do the trilinear interpolation [7].”
Virtual and Real World Adaptation for Pedestrian Detection (2014, background): “However, how to collect data on the internet is a non-trivial question that opens a new research area [6], [8] involving ethical questions too [9] since human paid work is at the core.”
Discriminative Decorrelation for Clustering and Classification (2012, result): “This is slightly lower than the SVM performance, but nearly equivalent to the original performance of [18].”
Pedestrian detection: A benchmark (2009, methodology): “All approaches use sliding windows and NMS (all except LatSvm use kernel density estimation for NMS, as proposed in [4]).”
Monocular Pedestrian Detection: Survey and Experiments (2009, methodology): “use kernel-based voting for position and scale of detected bounding boxes [10].”
Video-Based Human Behavior Understanding: A Survey (2013, methodology): “A related approach is also used by Dalal [62] and Bouchirika et al. [65], combining HOGs with motion and gait patterns.”
Multi-cue onboard pedestrian detection (2009, methodology): “In contrast to [4] we can clearly show a signiﬁcant perfor-Figure mance gain using motion features.”
Non-maximum Suppression for Object Detection by Passing Messages Between Windows (2014, background): “For instance, mean-shift for a kernel density estimation [15] and mixtures of scale-sensitive Gaussians [16] have been proposed.”
Human detection from images and videos: A survey (2016, methodology): “4: From left to right: HOG descriptor and using HOG to encode a local region of a human sample (from [21]).”
Pedestrian Detection via Classification on Riemannian Manifolds (2008, unknown): “Ç”