Multi-cue onboard pedestrian detection
The DOI matches its Crossref record; 36 of the 37 references it cites were found in a registry or an index.
Abstract
Various powerful people detection methods exist. Surprisingly, most approaches rely on static image features only despite the obvious potential of motion information for people detection. This paper systematically evaluates different features and classifiers in a sliding-window framework. First, our experiments indicate that incorporating motion information improves detection performance significantly. Second, the combination of multiple and complementary feature types can also help improve performance. And third, the choice of the classifier-feature combination and several implementation details are crucial to reach best performance. In contrast to many recent papers experimental results are reported for four different datasets rather than using a single one. Three of them are taken from the literature allowing for direct comparison. The fourth dataset is newly recorded using an onboard camera driving through urban environment. Consequently this dataset is more realistic and more challenging than any currently available dataset.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
Histograms of Oriented Gradients for Human Detection
Dalal, N., & Triggs, B. (2005) · 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) · doi:10.1109/cvpr.2005.177 · Cited by 33,584 · cited 5×
“Dalal et al. [6] enrich their static feature descriptor [ 5 ] with internal motion histograms to improve detection.”
BackgroundMethodResultInfluentialThe Ecological Approach to Visual Perception
Bornstein, M. H., & Gibson, J. J. (1980) · The Journal of Aesthetics and Art Criticism · doi:10.2307/429816 · Cited by 22,935 · cited 1×
“While psychologists and neuroscientists argue that motion is an important cue for human perception [17] only few computer vision object detectors (e.g. [30, 6]) exploit this fact.”
BackgroundRobust Real-Time Face Detection
Viola, P., & Jones, M. J. (2004) · International Journal of Computer Vision · doi:10.1023/b:visi.0000013087.49260.fb · Cited by 14,097
Friedman, J., Hastie, T., & Tibshirani, R. (2000) · The Annals of Statistics · doi:10.1214/aos/1016218223 · Cited by 6,884 · cited 1×
“AdaBoost Contrary to SVMs, boosting algorithms [14] optimize the classification error on the training samples iteratively.”
MethodA discriminatively trained, multiscale, deformable part model
Felzenszwalb, P., McAllester, D., & Ramanan, D. (2008) · 2008 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2008.4587597 · Cited by 2,867 · cited 2×
“Popular classifiers are SVMs [26, 5, 13, 20, 19] or boosting [31, 7, 33, 23].”
BackgroundDetecting Pedestrians Using Patterns of Motion and Appearance
Viola, P., Jones, M. J., & Snow, D. (2005) · International Journal of Computer Vision · doi:10.1007/s11263-005-6644-8 · Cited by 2,079
A Duality Based Approach for Realtime TV-L 1 Optical Flow
Zach, C., Pock, T., & Bischof, H. (2007) · Lecture Notes in Computer Science · doi:10.1007/978-3-540-74936-3_22 · Cited by 1,658 · cited 2×
“de Objectives Detect pedestrians from a moving platform • Exploit motion information • Leverage complementarity of features • Evaluate different classifiers • New datasets with image pairs Features • HOG [1] 8× 8 pixel cells, 2× 2 blocks 9-bin histograms, unsigned gradients • Haar wavelets [2] 32 and 16 pixel masks ho…”
BackgroundHuman Detection Using Oriented Histograms of Flow and Appearance
Dalal, N., Triggs, B., & Schmid, C. (2006) · Lecture Notes in Computer Science · doi:10.1007/11744047_33 · Cited by 1,652 · cited 7×
“In contrast, [ 6 ] proposed motion features that are ‐ at least in principle ‐ applicable to onboard sequences.”
BackgroundMethodResultInfluentialPedestrian detection: A benchmark
Dollar, P., Wojek, C., Schiele, B., & Perona, P. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206631 · Cited by 1,313 · cited 3×
“Doll´ ar et al. [ 8 ] report results for various recent approaches on a new challenging onboard dataset with improved evaluation metrics.”
BackgroundInfluentialSelf-citationA Trainable System for Object Detection
Papageorgiou, C., & Poggio, T. (2000) · International Journal of Computer Vision · doi:10.1023/a:1008162616689 · Cited by 1,289 · cited 3×
“Haar Haar wavelets have been introduced by Papageor-giou&Poggio [22] for people detection.”
BackgroundMethodInfluential
Research works printed but not linked: 1
- maji, 2008, Classification using intersection kernel SVMs is efficient, CVPR
Cited by
Items 201–300 of 370
Adapting a Pedestrian Detector by Boosting LDA Exemplar Classifiers
Xu, J., Vázquez, D., Ramos, S., López, A. M., & Ponsa, D. (2013) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2013.104 · Cited by 12
“The considered real-world datasets include INRIA [3] and ETH [17].”
BackgroundMethodSemi-Supervised Human Detection via Region Proposal Networks Aided by Verification
Wu, S., Wu, W., Lei, S., Lin, S., Li, R., Yu, Z., & Wong, H. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2944306 · Cited by 12
“INRIA [1], ETH [2], TUD-Brussels [3], Daimler [4],”
High-performance and deep pedestrian detection based on estimation of different parts
Saeidi, M., & Ahmadi, A. (2020) · The Journal of Supercomputing · doi:10.1007/s11227-020-03345-4 · Cited by 13
“The most popularly datasets include Caltech-USA [27], CityPersons [69], ETH [72], TUD-Brussels [73], INRIA [74], KITTI [75], and Daimler [76].”
Spatial-Temporal Granularity-Tunable Gradients Partition (STGGP) Descriptors for Human Detection
Liu, Y., Shan, S., Chen, X., Heikkilä, J., Gao, W., & Pietikäinen, M. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-15549-9_24 · Cited by 14
“In the third experiment, the STGGP detector is evaluated against the state of the art methods [29,25,3].”
MethodResultOut of sight: a toolkit for tracking occluded human joint positions
Wu, C.-J., Quigley, A. J., & Harris‐Birtill, D. (2016) · Personal and Ubiquitous Computing · doi:10.1007/s00779-016-0997-6 · Cited by 12
Bouafia, Y., Guezouli, L., & Lakhlef, H. (2022) · Iranian Journal of Science and Technology Transactions of Electrical Engineering · doi:10.1007/s40998-022-00512-6 · Cited by 12
“A study on a combination of features for pedestrian detection was done by Wojek and Schiele in (Wojek et al. 2009b).”
MethodDiscriminative Weighted Sparse Partial Least Squares for Human Detection
Li, Q., Yan, Y., & Wang, H. (2015) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2015.2495342 · Cited by 11
Robust pedestrian detection under deformation using simple boosted features
Kim, H.-K., & Kim, D. (2017) · Image and Vision Computing · doi:10.1016/j.imavis.2017.02.007 · Cited by 12
Pedestrian detection using HOG, LUV and optical flow as features with AdaBoost as classifier
Rabia, R., Shahid, A. R., Ziauddin, S., & Safi, A. (2016) · International Conference on Image Processing Theory, Tools and Applications (IPTA) · doi:10.1109/ipta.2016.7821024 · Cited by 13
“Wojek et al. [7] used HOG, Haar wavelets, and Support Vector Machine for pedestrian detection.”
MethodSOMPT22: A Surveillance Oriented Multi-pedestrian Tracking Dataset
Şimşek, F. E., Çığla, C., & Kayabol, K. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-25072-9_44 · Cited by 12
Joint global–local information pedestrian detection algorithm for outdoor video surveillance
Hu, H., Zhang, X., Zhang, W., & Li, B. H. (2014) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2014.11.009 · Cited by 10
Iaboni, C., Kelly, T., & Abichandani, P. (2025) · International Journal of Computer Vision · doi:10.1007/s11263-025-02418-2 · Cited by 12
A corpus for benchmarking of people detection algorithms
García‐Martín, Á., Martínez, J. M., & Bescós, J. (2011) · Pattern Recognition Letters · doi:10.1016/j.patrec.2011.09.038 · Cited by 11
“MotionPairs[4] Color pair-images 1310-pos Bounding box”
BackgroundInfluentialOn the purity of training and testing data for learning: The case of pedestrian detection
Taiana, M., Nascimento, J. C., & Bernardino, A. J. M. (2014) · Neurocomputing · doi:10.1016/j.neucom.2014.09.055 · Cited by 13
“The TUD-MotionPairs/TUD-Brussels data set [30] (TUD) and the Caltech pedestrian data set [12] were introduced in 2009 and contain sequences of images taken in automotive scenarios.”
MethodA Survey of Pedestrian Detection in Video
Solichin, A., Harjoko, A., & Putro, A. E. (2014) · International Journal of Advanced Computer Science and Applications · doi:10.14569/ijacsa.2014.051007 · Cited by 12
“TUDBrussels [43] 1776 218 1092 2009 [39], [44]”
Local Co-Occurrence Selection via Partial Least Squares for Pedestrian Detection
Li, Q., Wang, H., Yan, Y., Li, B., & Chen, C. W. (2016) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2016.2612705 · Cited by 11
“Most of the following experiments are performed on the INRIA pedestrian data set by using windows of 64×128 pixels except for the pedestrian detection results on the Caltech [31] and TUD-Brussels [32] pedestrian data sets.”
MethodInfluentialHuman head detection using Histograms of Oriented optical flow in low quality videos with occlusion
Hsu, F., Gubbi, J., & Palaniswami, M. S. (2013) · International Conference on Signal Processing and Communication Systems · doi:10.1109/icspcs.2013.6723967 · Cited by 11
“Furthermore, we propose to use histogram of oriented optical flow (HOOF) [24][4] as the feature.”
BackgroundMethodA Survey of Data Augmentation Techniques for Traffic Visual Elements
Yang, M., Sheng, E. L., Yew, W. K., Deng, S., & Kiong, T. S. (2025) · Sensors · doi:10.3390/s25216672 · Cited by 11
Pedestrian detection based on improved HOG feature and robust adaptive boosting algorithm
-, J. W., Yang, S., & Zhang, L. (2011) · International Congress on Image and Signal Processing (CISP) · doi:10.1109/cisp.2011.6100439 · Cited by 10
“[4] combine multiple types of features and motion information for pedestrian detection.”
BackgroundYano, S., Gu, Y., & Kamijo, S. (2014) · International Journal of Intelligent Transportation Systems Research · doi:10.1007/s13177-014-0103-2 · Cited by 10
Human Motion Capture Using Data Fusion of Multiple Skeleton Data
Masse, J.-T., Lerasle, F., Devy, M., Monin, A., Lefèbvre, O., & Mas, S. (2013) · Lecture notes in computer science · doi:10.1007/978-3-319-02895-8_12 · Cited by 9
“Keywords: Human Posture Reconstruction, Motion Capture, Data Fusion, Delayed Logic, Kalman Filter, Kinect.”
MethodFast pedestrian detection using a modified WLD detector in salient region
Lian, G., Lai, J., & Yang, Y. (2011) · International Conference on System Science and Engineering · doi:10.1109/icsse.2011.5961966 · Cited by 10
“[17] adds Haar wavelets, that is, the feature combination of HOG, optical flow and Haar wavelets is employed.”
BackgroundMethodHierarchical online domain adaptation of deformable part-based models
Xu, J., Vázquez, D., Mikolajczyk, K., & López, A. M. (2016) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2016.7487769 · Cited by 11
“For the target domain, we use three sequences from ETH dataset [26], namely ’BAHNHOF’, ’JELMOLI’, ’SUNNY DAY’, and denoted by ETH0, ETH1, ETH2 respectively.”
MethodRoad and Railway Smart Mobility: A High-Definition Ground Truth Hybrid Dataset
Khemmar, R., Mauri, A., Dulompont, C., Gajula, J., Vauchey, V., Haddad, M., & Boutteau, R. (2022) · Sensors · doi:10.3390/s22103922 · Cited by 12
“INRIA dataset [32], TUD-MotionPairs [33], and NightOwls [34].”
MethodHigh-Throughput Fixed-Point Object Detection on FPGAs
Ma, X., Najjar, W. A., & Roy-Chowdhury, A. (2014) · IEEE International Symposium on Field-Programmable Custom Computing Machines (FCCM) · doi:10.1109/fccm.2014.40 · Cited by 10
Pedestrain detection from motion
Kılıçarslan, M., Zheng, J. Y., & Raptis, K. (2016) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr.2016.7899907 · Cited by 10
Pandey, P., & Aghav, J. V. (2020) · Lecture notes in networks and systems · doi:10.1007/978-981-15-0694-9_27 · Cited by 9
Two-Stage Pedestrian Detection Model Using a New Classification Head for Domain Generalization
Schulz, D., & Perez, C. A. (2023) · Sensors · doi:10.3390/s23239380 · Cited by 8
“Currently, there are many benchmarks for assessing pedestrian detection; for example, Daimler [82], INRIA [39], ETH [83], TUDBrussels [84], and WiderPedestrian [85].”
BackgroundCheng, K., Khokhar, M. S., Liu, Q., Tahir, R., & Li, M. (2019) · IEEE Access · doi:10.1109/access.2019.2951164 · Cited by 9
“It contains 1063 frames from the datasets of Caltech [56], ETHZ [57], TUDBrussels [58], INRIA [59], Caviar [60].”
MethodTwo-stream small-scale pedestrian detection network with feature aggregation for drone-view videos
Xie, H., & Shin, H. (2021) · Multidimensional Systems and Signal Processing · doi:10.1007/s11045-021-00764-1 · Cited by 8
“In another popular dataset, the INRIA pedestrian dataset (Wojek et al. 2009), a 5% miss rate was reported with the method proposed in (Lin et al. 2018).”
MethodKatsamenis, I., Bakalos, N., Lappas, A. S., Protopapadakis, E. E., Montoliu, C. M.-P., Doulamis, A. D., Doulamis, N. D., Rallis, I., & Kalogeras, D. (2025) · Sensors · doi:10.3390/s25216653 · Cited by 9
Spatiotemporal Stacked Sequential Learning for Pedestrian Detection
González, A., Vázquez, D., Ramos, S., López, A. M., & Amores, J. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-19390-8_1 · Cited by 8
“, appearance + motion [32], appearance+depth+motion [12]), others have focused on the pedestrian model ( e.”
BackgroundFast Pedestrian Detection via Random Projection Features with Shape Prior
Zhao, Y., Yuan, Z., Chen, D., Lyu, J., & Liu, T. (2017) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2017.112 · Cited by 7
“These methods, however, come at significantly computational expense for training and testing due to the huge feature pools [29].”
BackgroundViS-HuD: Using Visual Saliency to Improve Human Detection with Convolutional Neural Networks
Gajjar, V., Khandhediya, Y., Gurnani, A., Mavani, V., & Raval, M. S. (2018) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2018.00256 · Cited by 7
“esults on the challenging achieve a competitive ark [30].”
BackgroundRobustness Analysis of Pedestrian Detectors for Surveillance
Fang, Y., Ding, G., Yuan, Y., Lin, W., & Liu, H. (2018) · IEEE Access · doi:10.1109/access.2018.2840329 · Cited by 7
Scan window based pedestrian recognition methods improvement by search space and scale reduction
Brehar, R. D., & Nedevschi, S. (2014) · IEEE Intelligent Vehicles Symposium · doi:10.1109/ivs.2014.6856571 · Cited by 7
“The reduction factor of the algorithm is further refined and decreased by applying a series of filters that successively eliminate part of the scan windows.”
MethodPedestrian Detection by Feature Selected Self-Similarity Features
Fu, X., Yu, R., Zhang, W., Li, F., & Shao, S. (2018) · IEEE Access · doi:10.1109/access.2018.2803160 · Cited by 7
“Apart from Caltech training VOLUME 6, 2018 14233 set, F-DNN+SS uses ImageNet [68], Microsoft COCO [69], Cityscape [70], ETH [4] and TudBrussels [5] for training, while we only use Caltech training set.”
BackgroundMethodMarathe, A., Walambe, R. A., & Kotecha, K. V. (2021) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw54120.2021.00398 · Cited by 15
“INRIA [6] contains pedestrians in bias-free poses in diverse settings, ETH [10] and TUD-Brussels [32] are midsized video datasets and ETH [10] and KITTI [14] provide stereo information.”
BackgroundA Monocular Human Detection System Based on EOH and Oriented LBP Features
Ma, Y., Chen, X., Liu, J., & Chen, G. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-24028-7_51 · Cited by 7
Fast visual people tracking using a feature-based people detector
Königs, A., & Schulz, D. (2011) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2011.6094902 · Cited by 7
“Positive training samples were drawn from TUD-Pedestrian ([10]), INRIA Person ([5]), TUD-Brussels ([ 21 ]), and MIT Pedestrian ([22]) datasets.”
MethodPedestrian Proposal and Refining Based on the Shared Pixel Differential Feature
Shen, J., Zuo, X., Zhu, L., Li, J., Yang, W., & Ling, H. (2018) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2018.2857814 · Cited by 7
“4) Results of the ETH Dataset: Using the same setting as the TUD-Brussel dataset, the model is trained on the INRIA dataset and evaluated in the ETH dataset.”
MethodInfluentialDeep Learning-Based Pedestrian Detection for Automated Driving: Achievements and Future Challenges
Karg, M. E., & Scharfenberger, C. (2019) · Studies in computational intelligence · doi:10.1007/978-3-030-31764-5_5 · Cited by 7
An FPGA-Accelerated Design for Deep Learning Pedestrian Detection in Self-Driving VehiclesPreprint
Moussawi, A., Haddad, K., & Chahine, A. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1809.05879 · Cited by 7
“In addition to Caltech dataset, we used both ETH pedestrian dataset [21] and TUD-Brussels datasets [22], that resulted in additional 4k images, and 14k annotations.”
MethodMiron, A. (2014) · theses.fr (ABES) · doi:10.70675/9e21af0az1310z4975za381zd77613dda859 · Cited by 10
“or near range [100-200] with 48% of the total number of BB.”
BackgroundMethodCo-occurrence flow for pedestrian detection
Maki, A., Seki, A., Watanabe, T., & Cipolla, R. (2011) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2011.6115837 · Cited by 6
“Although there have been few works which incorporate motion in pedestrian detection [16, 3, 21, 13], it was recently shown that additional use of motion features to HOG can enhance the performance for on-board sequences, in particular for pedestrians with side views which are of high importance in automotive safety ap…”
BackgroundMethodInfluentialWeakly Supervised Automatic Annotation of Pedestrian Bounding Boxes
Vázquez, D., Xu, J., Ramos, S., López, A. M., & Ponsa, D. (2013) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2013.107 · Cited by 6
Justifying the Importance of Color Cues in Object Detection: A Case Study on Pedestrian
Wang, Q., Pang, J., Qin, L., Jiang, S., & Huang, Q. (2012) · Springer eBooks · doi:10.1007/978-1-4614-3501-3_32 · Cited by 7
“Accurate detection would have immediate impacts on many real applications, such as intelligence surveillance [1] and driver assistance systems [2, 3].”
BackgroundMeasuring the Degree of Suitability of Edge Detection Operators Prior to an Application
Kesarwani, A., Purohit, K., Dalui, M., & Kisku, D. R. (2020) · IEEE Applied Signal Processing Conference (ASPCON) · doi:10.1109/aspcon49795.2020.9276678 · Cited by 6
“The proposed methodology has been evaluated with two publicly available databases, such as the Berkeley Segmentation Data Set and Benchmarks 300 (BSDS300) [17], and the Multi-cue boundary detection dataset [18].”
MethodOn experimenting with pedestrian classification using neural network
Rajesh, R., Rajeev, K., Gopakumar, V., Suchithra, K. P., & Lekhesh, V. P. (2011) · International Conference on Electronics Computer Technology · doi:10.1109/icectech.2011.5941967 · Cited by 6
“Different works which are directly or indirectly related to driver assistant systems are, but not limited to, people detection [7][9][12][19][20][26][27][28][29][33][34][35][36] [37][38][42][44][46][49] [50][51][52][53][56][57][60][61], people tracking [31], people detection and tracking [1][2][3] [5][7], pedestrian r…”
BackgroundAnalysis of Feature Fusion Based on HIK SVM and Its Application for Pedestrian Detection
Su, S., & Chen, S. (2013) · Abstract and Applied Analysis · doi:10.1155/2013/436062 · Cited by 6
“[13] combined the oriented histogram of flowwithHOGorHaar on an onboard camera setup and concluded that incorporating motion information considerably enhances detection performance.”
BackgroundRobust Pedestrian Classification Based on Hierarchical Kernel Sparse Representation
Sun, R., Zhang, G., Yan, X., & Gao, J. (2016) · Sensors · doi:10.3390/s16081296 · Cited by 6
“For different communities to benchmark and verify their pedestrian detection methods, many large-scale pedestrian data sets, including the Caltech [3], ETH [4], TUD-Brussels [5], Daimler [6], and INRIA [7] data sets, have been established and used as evaluation platforms.”
A Literature Review on Crowd Scene Analysis and Monitoring
Lamba, S., & Nain, N. (2016) · International Journal of Urban Design for Ubiquitous Computing · doi:10.21742/ijuduc.2016.4.2.02 · Cited by 6
Human Detection in Video over Large Viewpoint Changes
Duan, G., Ai, H., & Lao, S. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-19309-5_53 · Cited by 6
“We compare our approach with Intra-frame CF + Boost and Intra-frame CF + EMC-Boost approaches on PETS2007 dataset and our own collected videos, and also with [4] [20] [22] on ETHZ dataset.”
BackgroundMethodResultInfluentialRotative maximal pattern: A local coloring descriptor for object classification and recognition
Pang, J., Huang, J., Qin, L., Zhang, W., Qing, L., Huang, Q., & Yin, B. (2017) · Information Sciences · doi:10.1016/j.ins.2017.04.011 · Cited by 5
Adaptive algorithm selection, with applications in pedestrian detection
Zhang, S., Zhu, Q., & Roy–Chowdhury, A. K. (2016) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2016.7533064 · Cited by 5
“For instance, the datasets of CAVIAR [1] and TUD-Brussels [2] have been commonly used in the area of tracking.”
MethodInfluentialA Systematic Survey of ML Datasets for Prime CV Research Areas—Media and Metadata
Castro, H., Cardoso, J. S., & Andrade, M. T. (2021) · Data · doi:10.3390/data6020012 · Cited by 5
“The TUD dataset [89] was developed for research on pedestrian detection and tracking.”
BackgroundMethodPedestrian detection from non-smooth motion
Kılıçarslan, M., Zheng, J. Y., & Algarni, A. (2015) · IEEE Intelligent Vehicles Symposium · doi:10.1109/ivs.2015.7225732 · Cited by 5
An HOG-CT human detector with histogram-based search
Ding, J., Wang, Y., & Geng, W. (2011) · Multimedia Tools and Applications · doi:10.1007/s11042-011-0896-9 · Cited by 5
“Recently, researches have proved that single detectors with combining diverse existing features with different classifiers can improve the performance [31, 32, 34].”
BackgroundYi, Q., Chien, S. Y.-P., Li, L., Niu, W., Chen, Y., Good, D. H., Chen, C., & Sherony, R. (2018) · Journal of Intelligent and Connected Vehicles · doi:10.1108/jicv-02-2018-0005 · Cited by 5
Multiple Classifier Boosting and Tree-Structured Classifiers
Kim, T., & Cipolla, R. (2012) · Studies in computational intelligence · doi:10.1007/978-3-642-28661-2_7 · Cited by 8
“for pedestrian detection problems in [19].”
MethodPedestrian detection via contour fragments
Chen, D., Xia, S., & Zhou, Y. (2016) · Chinese Control Conference (CCC) · doi:10.1109/chicc.2016.7553986 · Cited by 5
“HOG + SVM framework has been one of the milestones in this field[9].”
BackgroundFaster robot perception using Salient Depth Partitioning
Chan, D. M., Taylor, A. M., & Riek, L. D. (2017) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2017.8206275 · Cited by 10
“However, existing datasets (such as PASCAL VOC [32], INRIA [1], ETH [41], TUD-Brussells [42], and Caltech [43]) do not contain spatiotemporal RGB-D images of pedestrians.”
BackgroundAdapting an object detector by considering the worst case: A conservative approach
Chen, G., Han, T. X., & Lao, S. (2011) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2011.5995362 · Cited by 4
“TUD-pedestrian dataset [21] is a new benchmark for pedestrian detection.”
BackgroundMethodHRBUST-LLPED: A Benchmark Dataset for Wearable Low-Light Pedestrian Detection
Li, T., Sun, G., Yu, L., & Zhou, K. (2023) · Micromachines · doi:10.3390/mi14122164 · Cited by 5
“In the following years, the INRIA [26], Daimler [27], and TUD-Brussels [28] datasets were successively introduced, with increased image resolutions of 640 × 480 and varying numbers of images, ranging from hundreds to tens of thousands.”
BackgroundIs pedestrian detection robust for surveillance?
Yuan, Y., Lin, W., & Fang, Y. (2015) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2015.7351308 · Cited by 4
“By addressing these content related challenges, various detection algorithms [6, 7, 8, 9, 2, 10, 11, 12] have been investigated to obtain higher detection accuracy.”
MethodInfluentialPost-processing approaches for improving people detection performance
García‐Martín, Á., & Martínez, J. M. (2014) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2014.09.010 · Cited by 6
“Global sequence performance is usually measured in terms of Precision-Recall (PR) curves [22], [37], [59].”
BackgroundA stereo cameras setup for pedestrian detection enhancement
Nakamura, A. T. M., Horita, L. R. T., & Grassi, V. (2017) · IEEE International Symposium on Robotics and Intelligent Sensors (IRIS) · doi:10.1109/iris.2017.8250147 · Cited by 4
“However, Wojek and Schiele [16] proposed a pedestrian detector with a combination of multiple features (Haar wavelet, HOG and Shape Context) that outperformed the best approach based on a singular feature.”
BackgroundRiečický, A., Madaras, M., Piovarči, M., & Ďurikovič, R. (2018) · International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP) · doi:10.5220/0006531100400047 · Cited by 4
“The trained probabilistic marker-less techniques (Wojek et al., 2009; Andriluka et al., 2010) that work on RGB images are not very precise in general.”
MethodDecade of Vision-Based Pedestrian Detection for Self-Driving: An Experimental Survey and Evaluation
Zhang, Y., Qi, Y., Liu, J., & Wang, Y. (2018) · SAE technical papers on CD-ROM/SAE technical paper series · doi:10.4271/2018-01-1603 · Cited by 4
Deep Learning for Object Detection: A Survey
Kotha, M. K., & Pavan, K. K. (2022) · Advanced technologies and societal change · doi:10.1007/978-981-19-4044-6_8 · Cited by 4
Integrating Perception and Cognition for AGI
Kurup, U., Lebiere, C. J., & Stentz, A. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-22887-2_11 · Cited by 4
“Approaches that deal with errors in perceptual processing are usually ad hoc and based on specialized heuristics that take advantage of the domain of interest [ 2 ].”
BackgroundHead detection using motion features and multi level pyramid architecture
Hsu, F., Gubbi, J., & Palaniswami, M. S. (2015) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2015.04.007 · Cited by 4
“Due to the complexity of the problem, it is clear that a single feature like HOG or HOOF, will prove to be insu ffi cient for describing the object [3].”
InfluentialExploiting feature dynamics for active object recognition
Robbel, P., & Roy, D. K. (2010) · International Conference on Control, Automation, Robotics and Vision (ICARCV) · doi:10.1109/icarcv.2010.5707373 · Cited by 5
“The usefulness of motion features has recently been demonstrated for pedestrian detection in [20].”
BackgroundNighttime FIR Pedestrian Detection Benchmark Dataset for ADAS
Xu, Z., Zhuang, J., Liu, Q., Zhou, J., & Peng, S. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-03341-5_27 · Cited by 5
Two-stage Deep Neural Network for General Object Detection
Linh, T. D., & Arai, M. (2019) · Journal of Information Processing · doi:10.2197/ipsjjip.27.268 · Cited by 5
Efficient Online Spatio-Temporal Filtering for Video Event Detection
Yan, X., Yuan, J., & Liang, H. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-16178-5_54 · Cited by 3
“Following [35], a walking pedestrian detector was previously trained by SVM with TUD-Motion Pairs [40] Dataset, while features of HOG [9], IMHd2, and Self-Similarity [37] are combined in the training step.”
MethodYamauchi, Y., & Fujiyoshi, H. (2011) · Asian Conference on Pattern Recognition · doi:10.1109/acpr.2011.6166556 · Cited by 3
“Most of the human detection methods proposed in recent years are based on a consideration of how to capture feature that are suitable for distinguishing humans, such as feature based on the shape of humans [4],[5], feature based on human motion [1]-[3],[6], and feature based on color information [7],[8].”
BackgroundRamirez-Martinez, M., Sanchez-Fernandez, F., Brunet, P., Senouci, S., & Bourennane, E. (2017) · International Conference on ReConFigurable Computing and FPGAs (ReConFig) · doi:10.1109/reconfig.2017.8279787 · Cited by 3
“We provide a comparison of our approach in different configurations in a database, we tested 6 different video sequences of the Marx Plank Institute computing department [15], [16].”
MethodLarge-Margin Learning of Compact Binary Image Encodings
Paisitkriangkrai, S., Shen, C., & van den Hengel, A. (2014) · IEEE Transactions on Image Processing · doi:10.1109/tip.2014.2337759 · Cited by 3
“In the next experiment, we apply C-BID to the baseline pedestrian detector of [52] using INRIA human data sets [52] and TUD-Brussels data sets [53].”
MethodMacrofeature layout selection for pedestrian localization and its acceleration using GPU
Nam, W., Han, B., & Han, J. H. (2013) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2013.10.011 · Cited by 3
Human detection using local shape and Non-Redundant binary patterns
Nguyen, D. T., Li, W., & Ogunbona, P. (2010) · International Conference on Control, Automation, Robotics and Vision (ICARCV) · doi:10.1109/icarcv.2010.5707303 · Cited by 3
“A study on features and classifiers selection for human detection was conducted recently by Wojek et al. in [19].”
BackgroundIsotropic Granularity-tunable gradients partition (IGGP) descriptors for human detection
Liu, Y., & Heikkila, J. (2010) · British Machine Vision Conference (BMVC) · doi:10.5244/c.24.63 · Cited by 3
“Even through remarkable progress has been achieved [2, 3, 4, 5, 8, 10, 20, 22, 25, 28], finding the human is still considered as one of the hardest task for object detection.”
BackgroundNoise Robust Hard Example Mining for Human Detection with Efficient Depth-Thermal Fusion
Zhao, Z., Zhang, J., & Shan, S. (2020) · International Conference on Automatic Face and Gesture Recognition · doi:10.1109/fg47880.2020.00121 · Cited by 4
“Most of them focus on the pedestrian detection, such as the classical works including INRIA [5], TudBrussels [6] and Daimler [7], as well as nowadays wildly used pedestrian dataset including Caltech-USA[8], KITTI[9] and CityPersons[10].”
Datasets, Metrics, Benchmarks and Future Research in Autonomous Driving: A Review
Li, Y., Teng, S., Wu, Z., Wang, J., Liu, M., Xuanyuan, Z., & Chen, L. (2026) · IEEE/CAA Journal of Automatica Sinica · doi:10.1109/jas.2025.125957 · Cited by 2
Gradient-based region of interest selection for faster pedestrian detection
Varga, R., & Nedevschi, S. (2013) · IEEE International Conference on Intelligent Computer Communication and Processing (ICCP) · doi:10.1109/iccp.2013.6646098 · Cited by 3
“Works in this direction are: [14], [15], [16], [17].”
BackgroundSteerable second order intensity features for pedestrian detection
Tan, S. K., Cham, T., & Wu, J. (2015) · Asian Conference on Pattern Recognition · doi:10.1109/acpr.2015.7486516 · Cited by 2
“Having trained our detector on the INRIA dataset [13], we additionally tested on the following datasets: TUDBrussel [20], ETH video sequence BAHNHOF [5], PennFudan [18], Daimler Chryler [12] to show that it is robust to different lighting conditions and scenes (car view, street view, walking video, and black and white…”
MethodA Shallow–Deep Feature Fusion Method for Pedestrian Detection
Liu, D., Zang, K., & Shen, J. (2021) · Applied Sciences · doi:10.3390/app11199202 · Cited by 2
“TUD-Brussel [49]: The training set contains 1092 image pairs with 1776 annotated pedestrians and 192 image pairs in the negative set.”
Parallel Deep Learning Framework for Video Surveillance System
Karvande, M., Katkar, A., Koli, N., Joshi, A. D., & Sawant, S. T. (2021) · Advances in parallel computing · doi:10.3233/apc210191 · Cited by 2
“For human detection, images from TUD Brussels dataset have been trained.”
MethodAn Adaptive Hardware Architecture using Quantized HOG Features for Object Detection
Nguyen, N.-D., Bui, D.-H., Hussin, F. A., & Tran, X. (2022) · IEEE International Conference on IC Design and Technology (ICICDT) · doi:10.1109/icicdt56182.2022.9933068 · Cited by 2
“However, for easier explantation and evaluation, we chose the pedestrian as the object detection with two datasets, including the INRIA dataset [2] and the TUD dataset [10].”
BackgroundMethodBalasubramanian, P., Pathak, S., & Mittal, A. (2016) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2016.150 · Cited by 4
“HOG [4] was subsequently improved by incorporating colour features [17, 28], texture reasoning [29, 24], motion cues [30], modifying and extending to multiple scales [13], improving run-time efficiency [32], and being used as a low-level component for other features in [1, 11].”
MethodPedestrian Detection Using Deep Channel Features in Monocular Image Sequences
Liu, Z., He, Y., Xie, Y. F., Gu, H., Liu, C., & Pei, M. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-46675-0_67 · Cited by 3
A study on occluded pedestrian detection based on block-based features and ensemble classifier
Wu, B., & Qu, S. (2015) · Chinese Control Conference (CCC) · doi:10.1109/chicc.2015.7260367 · Cited by 2
Towards Reliable Real-time Person Detection
Serban, S.-T., Simha, S. M., Bathrinarayanan, V., Corvée, E., & Brémond, F. (2014) · International Conference on Computer Vision Theory and Applications · doi:10.5220/0004651302320239 · Cited by 2
“We implement a variant of the Adaboost classifier that uses soft cascades (C. Zhang, 2007) for lossless reduction of detection time.”
MethodZeng, B., Wang, G., Ruan, Z., Xing-gang, L., & Meng, L. (2012) · Optical Engineering · doi:10.1117/1.oe.51.7.077206 · Cited by 2
Pedestrian detection in outdoor images using color and gradients
Häselich, M., Klostermann, M., & Paulus, D. W. R. (2013) · European Conference on Mobile Robots · doi:10.1109/ecmr.2013.6698857 · Cited by 2
“The evaluation was performed on the INRIA (InriaTest) [1] and the TudBrussels (TudBrussels) [13] datasets.”
BackgroundMethodInfluentialPedestrian detection with geometric context from a single image
Jun-qiang, W., & Ma, H. (2011) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/2072298.2071980 · Cited by 2
“Our test set is the publicly available TUD-Brussels [9] dataset.”
MethodDetecting bipedal motion from correlated probabilistic trajectories
Maki, A., Perbet, F., Stenger, B., & Cipolla, R. (2013) · Pattern Recognition Letters · doi:10.1016/j.patrec.2012.12.019 · Cited by 2
“We used the TUD-MotionPairs dataset (Wojek et al., 2009) for training the model as suggested in (Walk et al., 2010), and the histogram intersection kernel SVM (Maji et al., 2008) for the classifier. parison for Person A, rsimple with IMHd2 is a little higher than that with409 the proposed feet…”
MethodPedestrian Detection Based on Sparse and Low-Rank Matrix Decomposition
Ke-yang, C., Qi-rong, M., & Yong-zhao, Z. (2013) · TELKOMNIKA Indonesian Journal of Electrical Engineering · doi:10.11591/telkomnika.v12i2.3859 · Cited by 2
“Several authors have reported improvements by combining multiple types of low-level features [18-20].”
BackgroundRecognizing Complex Human Activities via Crowd Context
Choi, W., & Savarese, S. (2012) · Augmented vision and reality · doi:10.1007/8612_2012_4 · Cited by 2
“propose a probabilistic framework to detect individuals by combining multiple detectors [51] and by explicitly reasoning about target self-occlusions [50].”
BackgroundMethodRecent Developments in Pedestrian Detection Using Deep Learning
Karthika, N. J., & Chandran, S. (2019) · International Conference on Computing, Communication, and Intelligent Systems (ICCCIS) · doi:10.1109/icccis48478.2019.8974538 · Cited by 3
“Early efforts include INRIA[29], ETH [32], and TUD-Brussels [31] datasets.”
Metadata
- Authors
- Wojek, C., Walk, S., & Schiele, B.
- Affiliations
- Technische Universität Darmstadt (DE)
- ORCID
- Schiele, Bernt
- Venue
- 2009 IEEE Conference on Computer Vision and Pattern Recognition
- Publisher
- IEEE
- Published
- 2009-06
- Pages
- 794-801
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W2127420331
- Cited by
- 350
- References
- 37
- Citations by year
- 2026: 2 · 2025: 5 · 2024: 1 · 2023: 8 · 2022: 13 · 2021: 21 · 2020: 24 · 2019: 25 · 2018: 27 · 2017: 30 · 2016: 24 · 2015: 30 · 2014: 29 · 2013: 37 · 2012: 19
- FWCI
- 20.14
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings (publishedVersion)
- Max Planck Digital Library (submittedVersion)
- MPG.PuRe (Max Planck Society) (submittedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Video Surveillance and Tracking Methods
- Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Keywords
- pedestrian detection, onboard camera, motion information, multi-cue fusion, feature combination, sliding window, classifier comparison, urban driving
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.