What Makes for Effective Detection Proposals?
Open accessThe DOI matches its Crossref record; 69 of the 69 references it cites were found in a registry or an index.
Abstract
Current top performing object detectors employ detection proposals to guide the search for objects, thereby avoiding exhaustive sliding window search across images. Despite the popularity and widespread use of detection proposals, it is unclear which trade-offs are made when using them during object detection. We provide an in-depth analysis of twelve proposal methods along with four baselines regarding proposal repeatability, ground truth annotation recall on PASCAL, ImageNet, and MS COCO, and their impact on DPM, R-CNN, and Fast R-CNN detection performance. Our analysis shows that for object detection improving proposal localisation accuracy is as important as improving recall. We introduce a novel metric, the average recall (AR), which rewards both high recall and good localisation and correlates surprisingly well with detection performance. Our findings show common strengths and weaknesses of existing methods, and provide insights and metrics for selecting and tuning proposal methods.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“It is noteworthy that the Recall-to-IoU metric is just loosely [19], [20], [21] related to the ultimate detection accuracy.”
From Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (2017) · cites this paper as [20]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In the case of sliding window based detectors, the imbalance between backgrounds and objects could be as extreme as 10 7 :1 [71].”
From Object Detection in 20 Years: A Survey (2023) · cites this paper as [71]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472 · cited 1×
“Indeed, there has been some very recent work in this space [60], [61], [62] that shows promising”
BackgroundFaster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2015) · arXiv · doi:10.48550/arxiv.1506.01497 · Cited by 70,812 · cited 1×
“Indeed, there has been some very recent work in this space [60], [61], [62] that shows promising”
BackgroundImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333
Microsoft COCO: Common Objects in Context
Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., & Zitnick, C. L. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10602-1_48 · Cited by 45,460
Self-citationRich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.81 · Cited by 32,684
Girshick, R. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.169 · Cited by 28,105
Robust 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
Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2015) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2389824 · Cited by 11,971
Object Detection with Discriminatively Trained Part-Based Models
Felzenszwalb, P. F., Girshick, R. B., McAllester, D., & Ramanan, D. (2010) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2009.167 · Cited by 9,853
The Pascal Visual Object Classes Challenge: A Retrospective
Everingham, M., Eslami, S. M. A., Van Gool, L., Williams, C. K. I., Winn, J., & Zisserman, A. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-014-0733-5 · Cited by 7,139 · cited 3×
“We test the generalisation of proposal methods by evaluating on datasets with many additional classes in §4.”
Method
Research works printed but not linked: 7
- [28] A. Humayun, F. Li, and J. M. Rehg, 'Rigor: Recycling inference in graph cuts for generating object regions,' in CVPR , 2014. -
- [22] P. Krähenbühl and V. Koltun, 'Geodesic object proposals,' in ECCV , 2014. -
- [51] P. Dollár, R. Appel, and W. Kienzle, 'Crosstalk cascades for framerate pedestrian detection,' in ECCV , 2012. -
- [59] K. Tang, A. Joulin, L.-J. Li, and L. Fei-Fei, 'Co-localization in realworld images,' in CVPR , 2014. -
- [57] S. Vicente, C. Rother, and V. Kolmogorov, 'Object cosegmentation,' in CVPR , 2011. -
- [54] N. Chavali, H. Agrawal, A. Mahendru, and D. Batra, 'Objectproposal evaluation protocol is 'gameable',' arXiv:1505.05836 , 2015. - · arXiv
- szegedy, 0, Scalable, high-quality object detection, arXiv 1412 1441
Cited by
Items 201–300 of 764
Airport Detection Using End-to-End Convolutional Neural Network with Hard Example Mining
Cai, B., Jiang, Z., Zhang, H., Zhao, D., & Yao, Y. (2017) · Remote Sensing · doi:10.3390/rs9111198 · Cited by 41
“In these experiments, we employ the precision-recall (PR) curve, average precision (AP) [13], IoU-Recall (IR) curve [34], and F1-measure [28] to quantitatively evaluate the performance of our HEM-CNN method.”
MethodObject Detection in Autonomous Driving Scenarios Based on an Improved Faster-RCNN
Zhou, Y., Wen, S., Wang, D., Mu, J., & Irampaye, R. (2021) · Applied Sciences · doi:10.3390/app112411630 · Cited by 41
“First, region proposals are extracted [2], and then the region proposals are classified and the position coordinates are corrected.”
BackgroundDetection and Recognition for Life State of Cell Cancer Using Two-Stage Cascade CNNs
Hu, H., Guan, Q., Chen, S., Ji, Z., & Lin, Y. (2017) · IEEE Transactions on Computational Biology and Bioinformatics · doi:10.1109/tcbb.2017.2780842 · Cited by 43
“Currently, the object proposal algorithms falls into two categories: grouping methods and window scoring methods [19].”
BackgroundMethodAn Enhanced Region Proposal Network for object detection using deep learning method
Chen, Y. P., Li, Y., & Wang, G. (2018) · PLoS ONE · doi:10.1371/journal.pone.0203897 · Cited by 42
To what extent do DNN-based image classification models make unreliable inferences?
Tian, Y., Ma, S., Wen, M., Liu, Y., Cheung, S.-C., & Zhang, X. (2021) · Empirical Software Engineering · doi:10.1007/s10664-021-09985-1 · Cited by 44
Motorcycle detection and classification in urban Scenarios using a model based on Faster R-CNN
Espinosa, J. E., Velastín, S. A., & Branch, J. W. (2018) · International Conference on Pattern Recognition Systems (ICPRS) · doi:10.1049/cp.2018.1292 · Cited by 42
Object instance detection with pruned Alexnet and extended training data
Wang, R., Xu, J., & Han, T. X. (2018) · Signal Processing Image Communication · doi:10.1016/j.image.2018.09.013 · Cited by 43
Multi-Scale Proposal Generation for Ship Detection in SAR Images
Liu, N., Cao, Z., Cui, Z., Pi, Y., & Dang, S. (2019) · Remote Sensing · doi:10.3390/rs11050526 · Cited by 39
“We refer the reader to [53] for a survey of object proposal generation.”
BackgroundGan, H., Xu, C., Hou, W., Guo, J., Liu, K., & Xue, Y. (2022) · Biosystems Engineering · doi:10.1016/j.biosystemseng.2022.03.005 · Cited by 39
Multiscale Convolutional Neural Networks for Hand Detection
Yan, S., Xia, Y., Smith, J. S., Lu, W., & Zhang, B. (2017) · Applied Computational Intelligence and Soft Computing · doi:10.1155/2017/9830641 · Cited by 39
“The recall versus IoU curve was applied as the main evaluation metric for the region proposal algorithm in [40].”
MethodAn Improved YOLOv8-Based Lightweight Attention Mechanism for Cross-Scale Feature Fusion
Liu, S., Shao, F., Chu, W., Dai, J., & Zhang, H. (2025) · Remote Sensing · doi:10.3390/rs17061044 · Cited by 39
Vehicle type classification and attribute prediction using multi-task RCNN
Zhuoqun, H., Xia, Y., & Zhang, B. (2016) · International Congress on Image and Signal Processing (CISP) · doi:10.1109/cisp-bmei.2016.7852774 · Cited by 40
“However, we applied an alternative one, namely EdgeBoxes algorithm due to a number of appeals [17-18].”
MethodClass-specific object proposals re-ranking for object detection in automatic driving
Zhong, Z., Lei, M., Cao, D., Jianping Fan, B., & Li, S. (2017) · Neurocomputing · doi:10.1016/j.neucom.2017.02.068 · Cited by 40
“In addition, object proposal methods can make detection easier by removing false positives [8].”
MethodTowards accurate tiny vehicle detection in complex scenes
Liu, W., Liao, S., & Hu, W. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.03.004 · Cited by 37
Jain, P. K., Gupta, S., Bhavsar, A. V., Nigam, A., & Sharma, N. (2020) · Medical & Biological Engineering & Computing · doi:10.1007/s11517-019-02099-3 · Cited by 38
Jiang, C., Wang, Z., Liang, H., & Tan, S. (2022) · IEEE Sensors Journal · doi:10.1109/jsen.2022.3155232 · Cited by 38
“…(CPMC) (C) [30], 3 Endres (E) [29] 2 , geodesic object proposal (GOP) [32] 3 , Rantalankila (R1) [34] 3 , Objectness (Obj) [18] 3 , BING [17] 3 , EB [38] 3 , Rahtu (Ru) [37] 3 , Superpixel (S) [16] 3 , Uniform (U) [16] 3 , Gaussian (G) [16] 3 , region proposal network (RPN) [8], 4 and CPN [35].”
BackgroundMethodInfluentialWu, L., Chen, A., Salama, P., Winfree, S., Dunn, K. W., & Delp, E. J. (2023) · Scientific Reports · doi:10.1038/s41598-023-36243-9 · Cited by 38
Object proposal by multi-branch hierarchical segmentation
Wang, C., Zhao, L., Liang, S., Zhang, L., Jia, J., & Wei, Y. (2015) · 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.2015.7299012 · Cited by 39
“1The contemporary work [17] also perform such an object proposal evaluation using the R-CNN detector.”
MethodYoussef, Y., & Elshenawy, M. (2021) · Transportation Research Record Journal of the Transportation Research Board · doi:10.1177/0361198121997833 · Cited by 39
“The anchor box sizes and ratios in our implementation were [16, 32, 64, 128, 256] and [1, 0.7071, 1.4142] respectively.”
BackgroundMethodInfluentialAn Introduction to Deep Convolutional Neural Nets for Computer Vision
Srinivas, S., Sarvadevabhatla, R. K., Mopuri, K. R., Prabhu, N., Kruthiventi, S. S. S., & Babu, R. V. (2017) · Elsevier eBooks · doi:10.1016/b978-0-12-810408-8.00003-1 · Cited by 43
Deep Neural Networks for Marine Debris Detection in Sonar ImagesPreprint
Valdenegro-Toro, M. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.05241 · Cited by 45
Gan, H., Li, S., Ou, M., Yang, X., Huang, B., Liu, K., & Xue, Y. (2021) · Computers and Electronics in Agriculture · doi:10.1016/j.compag.2021.106384 · Cited by 36
Automated identification of diverse Neotropical pollen samples using convolutional neural networks
Punyasena, S. W., Haselhorst, D. S., Kong, S., Fowlkes, C. C., & Moreno, J. E. (2022) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.13917 · Cited by 35
“For a summary metric, we aver - aged recalls over all the IoU thresholds to report the average recall (Hosang et al., 2016 ).”
BackgroundMethodInfluentialYao, X., Zhang, H., Wen, S., Shi, Z., & Jiang, Z. (2024) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2024.3504865 · Cited by 35
Scale-Aware Pixelwise Object Proposal Networks
Jie, Z., Liang, X., Feng, J., Lu, W. F., Tay, E. H. F., & Yan, S. (2016) · IEEE Transactions on Image Processing · doi:10.1109/tip.2016.2593342 · Cited by 38
“This is because AR summarizes proposal performance across IoU thresholds and correlates well with object detection performance [37].”
BackgroundLiu, L., Yan, R.-J., Maruvanchery, V., Kayacan, E., Chen, I., & Tiong, L. K. (2017) · International Journal of Advanced Robotic Systems · doi:10.1177/1729881417712620 · Cited by 36
Coarse-to-Fine Localization of Temporal Action Proposals
Long, F., Yao, T., Qiu, Z., Tian, X., Mei, T., & Luo, J. (2019) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2019.2943204 · Cited by 35
“for unseen action categories [15], [35], [36], [46].”
BackgroundFirearm Detection and Segmentation Using an Ensemble of Semantic Neural Networks
Egiazarov, A., Mavroeidis, V., Zennaro, F. M., & Vishi, K. (2019) · European Intelligence and Security Informatics Conference · doi:10.1109/eisic49498.2019.9108871 · Cited by 34
“The existing methods address the segmentation problem by reformulating it into a classification problem; first a classifier is trained, and then, during the detection process, filters are run on a number of areas of the input image using either the sliding window approach [10]–[12] or the region proposals approach [13…”
MethodVision-Based Deep Learning Algorithm for Detecting Potholes
Gajjar, K., van Niekerk, T., Wilm, T., & Mercorelli, P. (2022) · Journal of Physics Conference Series · doi:10.1088/1742-6596/2162/1/012019 · Cited by 34
“AR is essentially the recall averaged over all IoU [0.5, 1.0] and can be determined by the following equation [19]: 𝐴𝐴𝑅𝑅 = 2 � 𝑟𝑟𝑟𝑟𝑟𝑟𝑅𝑅𝑅𝑅𝑅𝑅 ( 𝑜𝑜 0.5 where o is IoU and recall(o) is the corresponding recall.”
InfluentialSpatio-Temporal Closed-Loop Object Detection
Galteri, L., Seidenari, L., Bertini, M., & Bimbo, A. D. (2017) · IEEE Transactions on Image Processing · doi:10.1109/tip.2017.2651367 · Cited by 38
“Oneata et al. [29] follow a similar approach, in principle, by segmenting individual frames into a superpixel graph, then computing supervoxels through temporal hierarchical clustering.”
G-CNN: Object Detection via Grid Convolutional Neural Network
Lu, Q., Liu, C., Jiang, Z., Men, A., & Yang, B. (2017) · IEEE Access · doi:10.1109/access.2017.2770178 · Cited by 36
“R-CNN [16] uses the Selective Search method [19] to extract region proposals [20]; then, it uses CNNs”
MethodFastMask: Segment Multi-scale Object Candidates in One Shot
Hu, H., Lan, S., Jiang, Y., Cao, Z., & Sha, F. (2017) · 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.2017.245 · Cited by 37
“As average recall correlates well with object proposal quality [15], we summarize Average Recall (AR) between IoU 0.”
Gabor-CNN for object detection based on small samples
Hu, X.-D., Wang, X., Meng, F., Hua, X., Yan, Y.-J., Li, Y., Huang, J., & Jiang, X. (2019) · Defence Technology · doi:10.1016/j.dt.2019.12.002 · Cited by 35
Wu, X., Duan, J., Yang, L., & Duan, S. (2024) · Computers in Industry · doi:10.1016/j.compind.2024.104146 · Cited by 35
Applying Detection Proposals to Visual Tracking for Scale and Aspect Ratio Adaptability
Huang, D., Luo, L., Chen, Z., Wen, M., & Zhang, C. (2016) · International Journal of Computer Vision · doi:10.1007/s11263-016-0974-6 · Cited by 33
“Detection proposals can not only avoid performing classification on numerous windows, but also improve detection performance by reducing false positives (Hosang et al. 2014, 2015).”
BackgroundMethodInfluentialFast Joint Object Detection and Viewpoint Estimation for Traffic Scene Understanding
Guindel, C., Martín, D., & Armingol, J. M. (2018) · IEEE Intelligent Transportation Systems Magazine · doi:10.1109/mits.2018.2867526 · Cited by 36
“Therefore, considerable research effort is currently devoted to attention mechanisms generating these proposal windows [7].”
BackgroundObject tracking and detection techniques under GANN threats: A systemic review
Jaberi, S. M. A., Patel, A., & AL-Masri, A. N. (2023) · Applied Soft Computing · doi:10.1016/j.asoc.2023.110224 · Cited by 31
Bousetouane, F., & Morris, B. (2016) · IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS) · doi:10.1109/avss.2016.7738076 · Cited by 31
“In contrast to sliding window strategy, this paradigm generates a few hundred to thousands of regions which are likely to contain target objects [8].”
BackgroundCell Tracking via Proposal Generation and SelectionPreprint
Akram, S. U., Kannala, J., Eklund, L., & Heikkilä, J. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1705.03386 · Cited by 36
“In recent years, object proposals have become very popular in general object detection and segmentation, with almost all leading detection and segmentation methods relying on object proposals [16].”
Zhang, Y., Fu, K., Sun, H., Sun, X., Zheng, X., & Wang, H. (2017) · Remote Sensing Letters · doi:10.1080/2150704x.2017.1378452 · Cited by 30
“In (Hosang et al. 2016), lots of approaches are evaluated.”
Fully Convolutional Network for Multiscale Temporal Action Proposals
Guo, D., Li, W., & Fang, X. (2018) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2018.2839534 · Cited by 31
“2) Metrics: To be consistent with previous works [9], [10], [42], we employed two common metrics to measure the quality of the temporal proposals and evaluate the model performances.”
MethodPedestrian and cyclist detection based on deep neural network fast R-CNN
Wang, K., & Zhou, W. (2019) · International Journal of Advanced Robotic Systems · doi:10.1177/1729881419829651 · Cited by 30
Pixel Objectness: Learning to Segment Generic Objects Automatically in Images and Videos
Xiong, B., Jain, S. D., & Grauman, K. L. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2865794 · Cited by 32
“Alternatively, object proposal methods learn to localize all objects in an image, regardless of their category [13], [14], [15], [16], [17], [18], [19].”
BackgroundNazir, A., Cheema, M. N., Sheng, B., Li, P., Li, H., Yang, P., Jung, Y., Qin, J., & Feng, D. D. (2020) · Journal of Biomedical Informatics · doi:10.1016/j.jbi.2020.103430 · Cited by 29
He, S., & Lau, R. W. H. (2015) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.40 · Cited by 34
Learning to Group: A Bottom-Up Framework for 3D Part Discovery in Unseen CategoriesPreprint
Luo, T., Mo, K., Huang, Z., Xu, J., Hu, S., Wang, L., & Su, H. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.06478 · Cited by 36
“Mean Recall is then defined as the average values of Rt’s where t varies from 0.5 to 0.95 with 0.05 as a step size.”
InfluentialEvaluation of Object Proposals and ConvNet Features for Landmark-based Visual Place Recognition
Hou, Y., Zhang, H., & Zhou, S. (2017) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-017-0735-y · Cited by 28
“The first 10 proposals have been surveyed in detail in [21] so we will not repeat their description in this paper.”
BackgroundMethodResultInfluentialAnchor Generation Optimization and Region of Interest Assignment for Vehicle Detection
Wang, Y., Liu, Z., & Deng, W. (2019) · Sensors · doi:10.3390/s19051089 · Cited by 28
“The average recall (AR) proposed in [34] is used to evaluate the accuracy of RPN.”
MethodA region-based convolutional network for nuclei detection and segmentation in microscopy images
Liang, H., Cheng, Z., Zhong, H., Qu, A., & Chen, L. (2021) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2021.103276 · Cited by 32
SSD with multi-scale feature fusion and attention mechanism
Liu, Q., Dong, L., Zeng, Z., Zhu, W., Zhu, Y., & Chen, M. (2023) · Scientific Reports · doi:10.1038/s41598-023-41373-1 · Cited by 28
Two-stage pooling of deep convolutional features for image retrieval
Zhi, T., Duan, L., Wang, Y., & Huang, T. (2016) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2016.7532802 · Cited by 32
“As analyzed in [22], among most recent object proposal algorithms, BING is with the lowest computation complexity (the whole detection process only takes ∼10ms in a single thread over a normal PC with 2.”
MethodWell Control Space Out: A Deep-Learning Approach for the Optimization of Drilling Safety Operations
Magana-Mora, A., Affleck, M., Ibrahim, M., Makowski, G., Kapoor, H., Otalvora, W. B. C., Jamea, M. A., Umairin, I. S., Zhan, G. D., & Gooneratne, C. P. (2021) · IEEE Access · doi:10.1109/access.2021.3082661 · Cited by 29
Valdenegro-Toro, M. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-46182-3_18 · Cited by 29
“Control over the number of proposals is a desirable property of an proposal algorithm [6].”
BackgroundDeep learning‐based vehicle detection with synthetic image data
Wang, Y., Deng, W., Liu, Z., & Wang, J. (2019) · IET Intelligent Transport Systems · doi:10.1049/iet-its.2018.5365 · Cited by 27
“Average recall (AR) proposed in [46] is employed to evaluate RPN performance.”
MethodMagnuska, Z., Theek, B., Darguzyte, M., Palmowski, M., Stickeler, E., Schulz, V., & Kiessling, F. M. (2022) · Cancers · doi:10.3390/cancers14020277 · Cited by 26
“, high recall and high IoU) are more reproducible [42].”
BackgroundMethodCharacter proposal network for robust text extraction
Zhang, S., Lin, M., Chen, T., Jin, L., & Lin, L. (2016) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2016.7472154 · Cited by 26
“This is clearly different from the work [14], which applied FCN for semantic segmentation.”
MethodReal-time person detection in low-resolution thermal infrared imagery with MSER and CNNs
Herrmann, C., Müller, T., Willersinn, D., & Beyerer, J. (2016) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2240940 · Cited by 29
Deep Attribute-preserving Metric Learning for Natural Language Object Retrieval
Li, J., Wei, Y., Liang, X., Zhao, F., Li, J., Xu, T., & Feng, J. (2017) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3123266.3123439 · Cited by 29
“5 , 1] IoU vs. the number of proposals, which summarizes proposal performance across IoU thresholds [13].”
BackgroundZoom Out-and-In Network with Recursive Training for Object ProposalPreprint
Li, H., Liu, Y., Ouyang, W., & Wang, X. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1702.05711 · Cited by 26
“al [19] that region proposals with high average recall correlates well with good performance of a detector.”
BackgroundObject Localization Based on Proposal Fusion
Tang, S., Li, Y., Deng, L., & Zhang, Y. (2017) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2017.2729786 · Cited by 25
“avoiding exhaustive sliding window search across images [47].”
BackgroundThe Middle Child Problem: Revisiting Parametric Min-Cut and Seeds for Object Proposals
Humayun, A., Li, F., & Rehg, J. M. (2015) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.187 · Cited by 27
AccLoc: Anchor-Free and two-stage detector for accurate object localization
Piao, Z., Wang, J., Tang, L., Zhao, B., & Wang, W. (2022) · Pattern Recognition · doi:10.1016/j.patcog.2022.108523 · Cited by 25
CED-Net: context-aware ear detection network for unconstrained images
Kamboj, A., Rani, R., Nigam, A., & Jha, R. R. (2020) · Pattern Analysis and Applications · doi:10.1007/s10044-020-00914-4 · Cited by 25
“[13], [17], [42] presented a method to measure the accuracy of the predicted proposal by model, and signifies the importance of IOU.”
MethodA Combined Object Detection Method With Application to Pedestrian Detection
Gao, F., Wang, C., & Li, C. (2020) · IEEE Access · doi:10.1109/access.2020.3031005 · Cited by 25
“Considering the fact that the proportion of the human body is similar, so a fixed scale sliding widow is used to extract candidate regions, which can reduce the miss detection rate with less consumption of calculation resources [33].”
MethodBoosting Convolutional Features for Robust Object ProposalsPreprint
Karianakis, N., Fuchs, T. J., & Soatto, S. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1503.06350 · Cited by 28
Suhail, K., & Brindha, D. (2021) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2021.102806 · Cited by 24
Adaptive Non-Maximum Suppression for improving performance of Rumex detection
Al-Badri, A. H., Ismail, N. A., Al-Dulaimi, K., Salman, G. A., & Salam, M. S. H. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.119634 · Cited by 28
Locating 3D Object Proposals: A Depth-Based Online Approach
Pahwa, R. S., Lu, J., Jiang, N., Ng, T.-T., & Do, M. N. (2016) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2016.2616143 · Cited by 24
Object detection with class aware region proposal network and focused attention objective
Tao, X., Gong, Y., Shi, W., & Cheng, D. (2018) · Pattern Recognition Letters · doi:10.1016/j.patrec.2018.09.025 · Cited by 24
Towards Detection of Abnormal Vehicle Behavior Using Traffic Cameras
Wang, C., Musaev, A., Sheinidashtegol, P., & Atkison, T. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-23551-2_9 · Cited by 24
Pixel ObjectnessPreprint
Jain, S. D., Xiong, B., & Grauman, K. L. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1701.05349 · Cited by 24
“Given a batch of images known to contain the same object category, they segment the object in each one [25], [26], [27], [28], [20], [29].”
BackgroundAttend Refine Repeat: Active Box Proposal Generation via In-Out Localization
Gidaris, S., & Komodakis, N. (2016) · Procedings of the British Machine Vision Conference 2016 · doi:10.5244/c.30.90 · Cited by 23
Joint object detection and viewpoint estimation using CNN features
Guindel, C., Martín, D., & Armingol, J. M. (2017) · IEEE International Conference on Vehicular Electronics and Safety (ICVES) · doi:10.1109/icves.2017.7991916 · Cited by 24
“Therefore, considerable research effort is currently devoted to attention mechanisms generating these proposal windows [9].”
BackgroundHow Stable Are Transferability Metrics Evaluations?
Agostinelli, A., Pándy, M., Uijlings, J., Mensink, T., & Ferrari, V. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19830-4_18 · Cited by 44
Delving deep into the imbalance of positive proposals in two-stage object detection
Ge, Z., Jie, Z., Huang, X., Li, C., & Yoshie, O. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.10.098 · Cited by 23
DINS: A Diverse Insulator Dataset for Object Detection and Instance Segmentation
Cui, B., Han, C., Yang, M., Ding, L., & Shuang, F. (2024) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2024.3417290 · Cited by 22
“…recallratesandisusedtomeasurehowgoodthetrainedmodel isineachclass. mAPistheAPofallclasses Average recall (AR) [24] is an indicator to measure whether the positioning of a model detection frame is accurate that is the average of all recalls with intersection…”
BackgroundRobust Object Detection in Aerial Imagery Based on Multi-Scale Detector and Soft Densely Connected
Zhang, M., Zhang, B., Liu, M., & Xin, M. (2020) · IEEE Access · doi:10.1109/access.2020.2994379 · Cited by 23
“The average recall (AR) metric is used to evaluate the performance of each approach [41].”
MethodTubelets: Unsupervised Action Proposals from Spatiotemporal Super-Voxels
Jain, M., van Gemert, J. C., Jeǵou, H., Bouthémy, P., & Snoek, C. G. M. (2017) · International Journal of Computer Vision · doi:10.1007/s11263-017-1023-9 · Cited by 23
“The small set of object proposals is known to result in both high recall and overlap with the ground-truth (Hosang et al., 2016).”
Airplane detection in remote sensing images based on Object Proposal
Luo, Q., & Shi, Z. (2016) · IEEE International Geoscience and Remote Sensing Symposium (IGARSS) · doi:10.1109/igarss.2016.7729355 · Cited by 22
“The evaluation framework we used is the standard for Object Proposals methods [9].”
BackgroundMethodA Graph-Based Vehicle Proposal Location and Detection Algorithm
Yuan, X., Su, S., & Chen, H. (2017) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2017.2676182 · Cited by 22
“Two general object proposal algorithms have emerged [25]: grouping method and window scoring method.”
MethodAttentionMask: Attentive, Efficient Object Proposal Generation Focusing on Small Objects
Wilms, C., & Frintrop, S. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-20890-5_43 · Cited by 26
Deep Fusion Feature Based Object Detection Method for High Resolution Optical Remote Sensing Images
Wang, E. K., Li, Y., Nie, Z., Yu, J., Liang, Z., Zhang, X., & Yiu, S. (2019) · Applied Sciences · doi:10.3390/app9061130 · Cited by 23
“The BING algorithm is based on the sliding window scoring mechanism with an accelerating optimization, and it is considered to be the fastest region proposal algorithm [25].”
MethodLocal Region Proposing for Frame-Based Vehicle Detection in Satellite Videos
Zhang, J., Jia, X., & Hu, J. (2019) · Remote Sensing · doi:10.3390/rs11202372 · Cited by 22
“Inspired by the systematic region proposal evaluation research [42], the proposed region proposal scheme was evaluated against Superpixels (SP) [39,42], Selective Search (SS) [26] and Region Proposal Network”
MethodListNet-based object proposals ranking
Liu, Y., Zhang, X., Zhu, X., Guan, Q., & Zhao, X. (2017) · Neurocomputing · doi:10.1016/j.neucom.2017.06.008 · Cited by 25
“Object proposal methods aim to find out a set of proposal regions or bounding boxes, typically from hundreds to a few thousands per image, that tend to contain meaningful objects, independent of their categories [4].”
BackgroundMethodInfluentialSampaio, A. F., Rosado, L., & Vasconcelos, M. J. M. (2021) · IEEE Access · doi:10.1109/access.2021.3126486 · Cited by 21
“0 range, for a fixed number of detections [47].”
BackgroundACF: An Armed CCTV Footage Dataset for Enhancing Weapon Detection
Hnoohom, N., Chotivatunyu, P., & Jitpattanakul, A. (2022) · Sensors · doi:10.3390/s22197158 · Cited by 24
“[72], enhancing proposal localization accuracy is as crucial as improving recall for object detection.”
BackgroundBenchmarking Self-Supervised Contrastive Learning Methods for Image-Based Plant Phenotyping
Ogidi, F., Eramian, M., & Stavness, I. (2023) · Plant Phenomics · doi:10.34133/plantphenomics.0037 · Cited by 23
“We evaluated the model’s performance using the average precision (AP) [76] and average recall (AR) [77] metrics as implemented in the torchmetrics library (https://github.com/Lightning-AI/metrics/ blob/v0.”
Object Proposal Generation With Fully Convolutional Networks
Jie, Z., Lu, W. F., Sakhavi, S., Wei, Y., Tay, E. H., & Yan, S. (2016) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2016.2576759 · Cited by 22
“the object proposal robustness (called repeatability in [11]) when faced with image perturbation.”
BackgroundMethodInfluentialDeepProposals: Hunting Objects and Actions by Cascading Deep Convolutional Layers
Ghodrati, A., Diba, A., Pedersoli, M., Tuytelaars, T., & Gool, L. V. (2017) · International Journal of Computer Vision · doi:10.1007/s11263-017-1006-x · Cited by 21
“As shown in Hosang et al (2015), for a good detection performance, not only the recall is important, but also a good alignment of the candidates is needed.”
BackgroundMethodInfluentialCo-localization with Category-Consistent Features and Geodesic Distance Propagation
Le, H., Yu, C.-P., Zelinsky, G. J., & Samaras, D. (2017) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw.2017.134 · Cited by 25
“By finding those object categorical features, their joint locations can act as a single-shot object detector.”
BackgroundAtrous Faster R-CNN for Small Scale Object Detection
Guan, T., & Zhu, H. (2017) · International Conference on Multimedia and Image Processing (ICMIP) · doi:10.1109/icmip.2017.37 · Cited by 26
Generating region proposals for histopathological whole slide image retrieval
Ma, Y., Jiang, Z., Zhang, H., Xie, F., Zheng, Y., Shi, H., Zhao, Y., & Shi, J. (2018) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2018.02.020 · Cited by 21
Object Detection Based on Multi-sensor Proposal Fusion in Maritime Environment
Farahnakian, F., Haghbayan, M.-H., Poikonen, J. K., Laurinen, M., Nevalainen, P., & Heikkonen, J. (2018) · International Conference on Machine Learning and Applications · doi:10.1109/icmla.2018.00158 · Cited by 24
“These methods can generally be divided into two main categories: grouping methods and window scoring methods [8].”
MethodImproving Tiny Vehicle Detection in Complex Scenes
Liu, W., Liao, S., Hu, W., Liang, X., & Zhang, Y. (2018) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme.2018.8486507 · Cited by 22
“Following [22], an object is recalled if any proposal has IoU higher than 0.”
Deep Functional Dictionaries: Learning Consistent Semantic Structures on 3D Models from FunctionsPreprint
Sung, M., Su, H., Yu, R., & Guibas, L. J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.09957 · Cited by 35
“3 As an evaluation metric, we use proposal recall [13], which measures the percentage (3)Wang et al.”
MethodDeep network aided by guiding network for pedestrian detection
Jung, S., & Hong, K.-S. (2017) · Pattern Recognition Letters · doi:10.1016/j.patrec.2017.02.018 · Cited by 20
Using Deep Learning to Identify Potential Roof Spaces for Solar Panels
House, D., Lech, M., & Stolar, M. N. (2018) · International Conference on Signal Processing and Communication Systems · doi:10.1109/icspcs.2018.8631725 · Cited by 21
“CNNs have been shown to be susceptible to different blurring and noise while compression has little effect [10], this process will likely help interpret results when using low resolution satellite images.”
BackgroundZhang, Z., Hua, Z., Tang, Y., Zhang, Y., Lu, W., & Dai, C. (2021) · Actuators · doi:10.3390/act10080170 · Cited by 20
“Average Precision (AP) is the percentage of the number of correctly identified objects in the total number of objects, and mAP (Mean Average Precision) is used as an index to evaluate the accuracy of detection of all categories [40].”
MethodHierarchical Image Segmentation Ensemble for Objectness in RGB-D Images
Wang, H., Huang, D., Jia, K., & Wang, Y. (2017) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2017.2776220 · Cited by 20
“The average recalls (AR) [19] for 10, 100, and 1000 proposals are also adopted to contrast previous deep model based approaches such as DeepMask [28] and SharpMask [32].”
BackgroundMethodA Block Object Detection Method Based on Feature Fusion Networks for Autonomous Vehicles
Meng, Q., Song, H., Li, G., Zhang, Y., & Zhang, X. (2019) · Complexity · doi:10.1155/2019/4042624 · Cited by 20
“Currently, datasets in the field of computer vision include ImageNet [7], COCO [8] and PASCAL VOC [9], etc. More-over, the dataset for autonomous driving primarily uses the KITTI [10] dataset or the Cityscapes [11] dataset.”
Background
Metadata
- Authors
- Hosang, J., Benenson, R., Dollar, P., & Schiele, B.
- Affiliations
- Max Planck Institute for Informatics (DE)
- ORCID
- Schiele, Bernt
- Venue
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2016-04-01
- Volume
- 38
- Issue
- 4
- Pages
- 814-830
- ISSN
- 0162-8828 (ISSN-L)
- 2160-9292
- OpenAlex
- W1958328135
- Cited by
- 727
- References
- 68
- Citations by year
- 2026: 9 · 2025: 27 · 2024: 34 · 2023: 36 · 2022: 57 · 2021: 58 · 2020: 63 · 2019: 95 · 2018: 93 · 2017: 126 · 2016: 98
- FWCI
- 100.45
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- MPG.PuRe (Max Planck Society) (submittedVersion, other-oa)
- MPG.PuRe (Max Planck Society) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Visual Attention and Saliency Detection
- Keywords
- object detection, PASCAL dataset, ImageNet, MS COCO, DPM, R-CNN, Fast R-CNN, sliding window search
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.