An HOG-LBP human detector with partial occlusion handling

By Tony X. Han, Shuicheng Yan. IEEE International Conference on Computer Vision, 2009

570 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/16404d38775e2e8ddcd28e2ca6e79cf8a82d0d9f

Citations in the corpus

• Filtered channel features for pedestrian detection (2015, unknown): “LBP) Instead of using the magnitude of local pixel differences, LBP uses the difference sign only as signal [41, 42, 29].”
• Modeling Occlusion by Discriminative AND-OR Structures (2013, unknown): “…estimate the visibilities of parts from image appearance, using assumptions that the visibility of a part is independent from other parts [30] [29] [13], is consistent with neighbouring parts [7] [9], or is consistent with its parent or child parts describing object appearance at different…”
• Data-driven 3D Voxel Patterns for object category recognition (2015, unknown): “In order to detect partially occluded objects, researchers have worked on training partial object detectors for visible parts of objects [38, 35, 15, 37, 41].”
• Automated identification of animal species in camera trap images (2013, unknown): “In the area of computer vision, for human detection [13], HOG and cLBP features are concatenated to obtain the final feature.”
• Efficient HOG human detection (2011, unknown): “As reported in [27], the superiority of the complete convolution method is exhibited when occlusion processing and LBP features are included.”
• A Multilevel Mixture-of-Experts Framework for Pedestrian Classification (2011, unknown): “and motion. While some approaches utilize combinations on the module level [2], [12], [13], [19], [37], [47], others integrate multiple information sources directly into the pattern classification step [5], [9], [40], [43], [44], [49], [51]‐[ 53 ], [56], [58].”
• Deep learning for visual understanding: A review (2016, unknown): “Sparse coding is one of the basic algorithms in deep learning, and it is more discriminative than the original hand-designed ones, i.e. HOG [132] and LBP [133].”
• Single-Pedestrian Detection Aided by Multi-pedestrian Detection (2013, unknown): “By integrating with our framework, 10 algorithms outperform LatSVM-V2 and 0 10 20 30 FPDW MultiFtr+Motion HOG HikSvm Shapelet ChnFtrs MultiFtr+CSS PoseInv Pls VJ HogLbp MultiFtr LatSvm−V1 LatSvm−V2 MultiResC 4 5 10 15 18 4 6 24 5 9 2 9 15 5 5 % 0 10 20 FPDW MultiFtr+Motion HOG HikSvm Shapelet ChnFtr”
• A Survey of Recent Advances in Face Detection (2010, unknown): “In [37] [119], LBP was applied for face detection tasks under a Bayesian and a boosting framework, respectively.”
• A survey on face detection in the wild: Past, present and future (2015, unknown): “Featuresforface/objectdetection. featuresand RotatedHaar-likefeatures[63] itsvariations Rectangularfeatureswithstructure[59, 64] Haar-likefeaturesonmotionfiltered image[65] Pixel-based Pixelpairs[78] features Controlpointset[79] Binarized Modifiedcensustransform[69] features LBPfeatures[71,72] Local”
• BRIEF: Computing a Local Binary Descriptor Very Fast (2012, unknown): “Extensions are numerous and include [25] , [26], [27], [28], [29], [30].”
• Efficient additive kernels via explicit feature maps (2010, unknown): “For instance, the miss rate at false positive per window rate (FPPW) 10 4 is 0.05 for the HOG descriptor from Felzenszwalb et al. [7] with the χ 2 3D approximated kernel, whereas Ott and Everingham [17] reports 0.05 integrating HOG with image segmentation and using a quadratic kernel, Wang et al. [ ”
• Detection Evolution with Multi-order Contextual Co-occurrence (2013, unknown): “0 th 1 st 1 st + H 0 th + 1 st 0 th + 1 st + H SURF LBP 45.5 46.0 46.7 46.8 47.2 44.7 45 Table 2: Mean AP (mAP) varies with the combination of different order context feature, where 0 th , 1 st , H respectively refers to 0 th , 1 st and high order descriptors.”
• Local binary features for texture classification: Taxonomy and experimental study (2017, unknown): “[97] combined Histogram of Gradients (HOG) with LBP, performing very well in human detection with partial occlusion handling.”
• New features and insights for pedestrian detection (2010, unknown): “HOG was recently enriched by Local Binary Patterns (LBP), showing a visible improvement over standard HOG on the INRIA Person data set [24].”
• Boosted local structured HOG-LBP for object localization (2011, unknown): “Detector models mainly include part based models [8] [9] [20] [27] and rigid template models [1] [24] [25].”