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 401–500 of 764
Improving Keypoint Matching Using a Landmark-Based Image Representation
Huang, X., Dai, Z., Chen, W., He, L., & Zhang, H. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8794420 · Cited by 9
“to [19].”
Fine-Grained Action Recognition by Motion Saliency and Mid-Level Patches
Liu, F., Zhao, L., Cheng, X., Dai, Q., Shi, X., & Qiao, J. (2020) · Applied Sciences · doi:10.3390/app10082811 · Cited by 9
“Furthermore, MCG (Multiscale Combinatorial Grouping) [43], Multibox [44], and other algorithms [45] provide alternative proposal methods.”
MethodDeeply Activated Salient Region for Instance Search
Xiao, H.-C., Zhao, W., Jie, L., Hong, Y.-G., & Ngo, C. (2022) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3510004 · Cited by 10
“Following [47], recall is adopted to measure the fraction of a detected region that overlaps with its ground-truth instance based on IoU threshold.”
MethodTracking objects using 3D object proposals
Pahwa, R. S., Ng, T.-T., & Do, M. N. (2017) · Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) · doi:10.1109/apsipa.2017.8282298 · Cited by 9
“Unfortunately, these algorithms have limited repeatability [14].”
Object Detection Using Deep Convolutional Neural Networks
Qian, H., Xu, J., & Zhou, J. (2018) · Chinese Automation Congress · doi:10.1109/cac.2018.8623808 · Cited by 10
“Frequently-used region proposal generation algorithms [17] include the grouping proposal methods like Selective Search (SS) and the window scoring proposal methods like Objectness.”
MethodLandmark Generation in Visual Place Recognition Using Multi-Scale Sliding Window for Robotics
Yang, B., Xu, X., Li, J., & Zhang, H. (2019) · Applied Sciences · doi:10.3390/app9153146 · Cited by 9
Military Object Detection Based on Optimal Gabor Filtering and Deep Feature Pyramid Network
Hu, X., Zhang, P., & Xiao, Y. (2019) · International Conference on Artificial Intelligence and Computer Science (AICS) · doi:10.1145/3349341.3349462 · Cited by 9
Dog Breed Classifier using Convolutional Neural Networks
Rishita, M. V. S., & Harris, T. A. (2018) · International Conference on Networking, Embedded and Wireless Systems (ICNEWS) · doi:10.1109/icnews.2018.8903980 · Cited by 9
Pedestrian Detection Based on Panoramic Depth Map Transformed from 3D-LiDAR Data
Chen, G., Mao, Z., Yi, H., Li, X., Bai, B., Liu, M., & Zhou, H. (2020) · Periodica Polytechnica Electrical Engineering and Computer Science · doi:10.3311/ppee.14960 · Cited by 12
“From [39], the RPN feature of Faster-RCNN [7] is not conducive to the detection of small objects, which results in the AP of Faster-RCNN [7] being significantly smaller than PVANET.”
ResultWhat Leads to Generalization of Object Proposals?
Wang, R., Mahajan, D. K., & Ramanathan, V. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-66096-3_32 · Cited by 13
Exploit Direction Information for Remote Ship Detection
Tan, Z., Zhang, Z., Xing, T., Huang, X., Gong, J., & Ma, J. (2021) · Remote Sensing · doi:10.3390/rs13112155 · Cited by 9
“Faster RCNN [24] replaced region proposal algorism [25] by region proposal networks (RPN), making the whole detector “end-to-end” and approximately real-time.”
BackgroundEnhancing representation learning by exploiting effective receptive fields for object detection
Wang, Q., Zhang, S., Yu, Q., Zhang, G., & Wang, H. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.01.020 · Cited by 9
Visual Detection of Unknown Objects in Video Games Using Qualitative Stability Analysis
Ge, X., Renz, J., & Zhang, P. (2015) · IEEE Transactions on Computational Intelligence and AI in Games · doi:10.1109/tciaig.2015.2506741 · Cited by 11
“Detection proposal methods (for a survey, see [13]) have been widely applied to object detection.”
MethodWeakly Supervised Text Attention Network for Generating Text Proposals in Scene Images
Li, R., En, M., Jian-qiang, L., & HaiBin, Z. (2017) · IAPR International Conference on Document Analysis and Recognition (ICDAR) · doi:10.1109/icdar.2017.61 · Cited by 11
“Evaluation protocol presented in [36] is used.”
Opportunistic named functions in disruption-tolerant emergency networks
Graubner, P., Lampe, P., Höchst, J., Baumgärtner, L., Mezini, M., & Freisleben, B. (2018) · ACM International Conference on Computing Frontiers (CF) · doi:10.1145/3203217.3203234 · Cited by 9
“Thus, as many faces as possible are detected quickly.”
BackgroundSalient object detection via boosting object-level distinctiveness and saliency refinement
Yan, X., Wang, Y., Song, Q., & Dai, K. (2017) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2017.06.013 · Cited by 8
RETR: Multi-View Radar Detection Transformer for Indoor Perception
Yataka, R., Cardace, A., Wang, P. P., Boufounos, P. T., & Takahashi, R. (2024) · neural information processing systems · doi:10.52202/079017-0625 · Cited by 23
MATHEMATICAL MODELING OF ORGANIZATIONAL SYSTEMS
Grachev, M. I., Burlov, V. G., & Gracheva, N. G. (2022) · H&ES Research · doi:10.36724/2409-5419-2022-14-5-14-20 · Cited by 8
Hand-drawn Electric Circuit Diagrams Recognition using Deep Learning
AlMughrabi, A., & Hiary, H. (2024) · doi:10.1109/it61232.2024.10475731 · Cited by 8
“…the bounding boxes; the authors [2] proposed the Region-based CNN (R-CNN) approach to bounding box object detection, and it is attending a manageable number of candidate object regions (i.e., proposals) [3], [4] and evaluate convolutional networks independently on each Region of Interest (RoI).”
BackgroundA Novel Structure Fusion Attention Model to Detect Architectural Distortion on Mammography
Ou, T.-W., Weng, T.-C., & Chang, R. (2024) · Journal of Imaging Informatics in Medicine · doi:10.1007/s10278-024-01085-y · Cited by 8
“The AR is thereby defined in a study of Hosang et al. [48] as where o represents the IoU threshold and Recall( o ) is the Recall at the corresponding IoU threshold for IoU ∈ [0.5, 1.0].”
InfluentialPedestrian detection based on Region Proposal Fusion
Wang, B., Tang, S., Zhao, R., Liu, W., & Cen, Y. (2015) · IEEE International Workshop on Multimedia Signal Processing (MMSP) · doi:10.1109/mmsp.2015.7340847 · Cited by 8
“To overcome the problem, region proposals are proposed [20].”
BackgroundMWNet: object detection network applicable for different weather conditions
Liu, P., Yuan, X., & Dai, X. (2019) · IET Intelligent Transport Systems · doi:10.1049/iet-its.2019.0023 · Cited by 8
Cascaded Sparse Spatial Bins for Efficient and Effective Generic Object Detection
Novotný, D., & Matas, J. (2015) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.137 · Cited by 8
“However their approach is not tuned for specific amounts of outputted proposals thus, when for instance using a very dense sampling of EdgeBoxes during the first stage of our SSPB cascade, the method of [17] would be comparable to running simple greedy NMS with threshold set to 0.”
BackgroundMethodResultInfluentialEnergy-Efficient Object Detection Using Semantic Decomposition
Panda, P., Venkataramani, S., Sengupta, A., Raghunathan, A., & Roy, K. (2017) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2017.2707077 · Cited by 10
“It is worth mentioning that our hierarchical classification methodology is complementary to the concept of cascading classifiers proposed in [3] and [4].”
MethodExploring to learn visual saliency: The RL-IAC approach
Craye, C., Lesort, T., Filliat, D., & Goudou, J.-F. (2018) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2018.11.012 · Cited by 8
A robust proposal generation method for text lines in natural scene images
Fan, K., & Baek, S. J. (2018) · Neurocomputing · doi:10.1016/j.neucom.2018.03.041 · Cited by 8
Hyper Feature Fusion Pyramid Network for Object Detection
Huang, S., Li, X., Jiang, Z., Guo, X., & Men, A. (2018) · IEEE International Conference on Multimedia & Expo Workshops (ICMEW) · doi:10.1109/icmew.2018.8551547 · Cited by 8
“Proposals are used to guide the search for objects to avoid exhaustive sliding window search across images, and many proposal algorithms have been proposed [18].”
BackgroundDetection Process of Ships in Aerial Imagery Using Two Convnets
Zerrouk, I., Moumen, Y., Khiati, W., Berrich, J., & Bouchentouf, T. (2019) · International Conference on Wireless Technologies, Embedded and Intelligent Systems (WITS) · doi:10.1109/wits.2019.8723734 · Cited by 8
“[7] Jan Hosang, Rodrigo Benenson, Piotr Dollár, and Bernt Schiele, “What makes for effective detection proposals?”, CoRR, Volume abs/1502.”
Background[Artificial intelligence empowers laboratory medicine in Industry 4.0].
Zhou, Q., Qi, S., Xiao, B., Li, Q., Sun, Z., & Li, L. (2020) · PubMed · doi:10.12122/j.issn.1673-4254.2020.02.23 · Cited by 8
Enhancing aphid detection framework based on ORB and convolutional neural networks
Pei, H., Liu, K., Xiaojing, Z., & Yahya, A. A. (2020) · Scientific Reports · doi:10.1038/s41598-020-75721-2 · Cited by 8
Dong, R., Jiao, L., Zhang, Y., Zhao, J. K., & Shen, W. (2021) · Remote Sensing · doi:10.3390/rs13173362 · Cited by 8
Face Recognition Technology Improving Using Convolutional Neural Networks
Miakshyn, O., Anufriiev, P., & Bashkov, Y. (2021) · IEEE International Conference on Advanced Trends in Information Theory (ATIT) · doi:10.1109/atit54053.2021.9678722 · Cited by 8
Automatic estimation of rice grain number based on a convolutional neural network
Deng, R., Qi, L., Pan, W., Wang, Z., Fu, D., & Yang, X. (2022) · Journal of the Optical Society of America A · doi:10.1364/josaa.459580 · Cited by 10
Nagrath, P., Thakur, N., Jain, R., Saini, D., Sharma, N., & Hemanth, D. J. (2022) · Internet of things · doi:10.1007/978-3-030-89554-9_2 · Cited by 11
Tree-based indexing for real-time ConvNet landmark-based visual place recognition
Hou, Y., Zhang, H., & Zhou, S. (2017) · International Journal of Advanced Robotic Systems · doi:10.1177/1729881416686951 · Cited by 8
“illumination, rotation, scaling, and so on).(37) Thanks to its good repeatability, impressive recognition accuracy in the presence of severe environmental changes has been achieved in the work by Süenderhauf et al.”
BackgroundSegmentation Mask Refinement Using Image Transformations
Nguyen, T. D., Shinya, A., Harada, T., & Thawonmas, R. (2017) · IEEE Access · doi:10.1109/access.2017.2772269 · Cited by 8
Bi-Connect Net for salient object detection
Jia, F., Wang, X., Guan, J., Liao, Q., Zhang, J., Li, H., & Qi, S. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.12.020 · Cited by 8
Fast Region Proposal Learning for Object Detection for RoboticsPreprint
Ceola, F., Maiettini, E., Pasquale, G., Rosasco, L., & Natale, L. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.12790 · Cited by 8
FashionFail: Addressing Failure Cases in Fashion Object Detection and Segmentation
Velioğlu, R., Chan, R., & Hammer, B. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10651287 · Cited by 7
“Thus, the weighted mean average recall [68] for the top k detections is computed as where the class weights w c for classes c ∈ { 1 , . . . , C } are the same as in Equation (2) and AR top k c is the average recall as in Equation (4) but restricted on class c .”
MethodKatzenmaier, M., Garnot, V. S. F., Wegner, J. D., & von Arx, G. (2025) · Frontiers in Plant Science · doi:10.3389/fpls.2025.1516635 · Cited by 8
A comprehensive study on object proposals methods for vehicle detection in aerial images
Sommer, L. W., Schuchert, T., & Beyerer, J. (2016) · IAPR Workshop on Pattern Recognition in Remote Sensing (PRRS) · doi:10.1109/prrs.2016.7867018 · Cited by 7
AI-based boulder detection in sonar data – bridging the gap from experimentation to application
Hinz, M., Westfeld, P., Feldens, P., Feldens, A., Themann, S., & Papenmeier, S. (2024) · The International Hydrographic Review · doi:10.58440/ihr-30-1-a08 · Cited by 7
Proposing regions from histopathological whole slide image for retrieval using Selective Search
Ma, Y., Jiang, Z., Zhang, H., Xie, F., Zheng, Y., & Shi, H. (2017) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi.2017.7950491 · Cited by 7
“In the review of various RP methods [3], we find that Selective Search (SS) [4] can reach high recall and output segment in small amount of proposal without annotation.”
MethodCnn Based Renormalization Method for Ship Detection in Vhr Remote Sensing Images
Wang, T., & Gu, Y. (2018) · IEEE International Geoscience and Remote Sensing Symposium (IGARSS) · doi:10.1109/igarss.2018.8518680 · Cited by 7
“The traditional method is Sliding window, Object proposals[2] (super pixel based or some other unsupervised segmentation methods) like R-CNN[3], Fast R-CNN[4], Faster R-CNN[5], and bounding box regression like You Only Look Once (YOLO)[6] and Single Shot MultiBox Detector (SSD)[7].”
MethodSiamese Tracking with Adaptive Template-Updating Strategy
Zheng, X., Luo, H., Hui, B., Chang, Z., & Ju, M. (2019) · Applied Sciences · doi:10.3390/app9183725 · Cited by 7
Li, H., Wang, C., Wang, J., & Zhang, Y. L. (2022) · Structural Health Monitoring · doi:10.1177/14759217221123485 · Cited by 7
“However, there are controversies about this standard in general object detection, 30,31 that the mAP scores sometimes could not indicate the visual performance of the detection reasonably.”
BackgroundAnalysis of the Impact of Rain on Perception in Automated Vehicle Applications
Brophy, T., Mullins, D., Parsi, A., Horgan, J., Ward, E., Denny, P., Eising, C., Deegan, B. M., Glavin, M., & Jones, E. (2025) · IEEE Open Journal of Vehicular Technology · doi:10.1109/ojvt.2025.3553718 · Cited by 7
“Average Recall (AR) is another metric commonly used to evaluate machine learning performance [36].”
Temporal objectness: Model-free learning of object proposals in video
Peng, L., & Qi, X. (2016) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2016.7533043 · Cited by 7
“Index Terms— Object Detection, Object Proposals, Video, Temporal Objectness, Model-Free Learning”
BackgroundMukherjee, P., & Lall, B. (2020) · Signal Processing Image Communication · doi:10.1016/j.image.2020.115890 · Cited by 7
“In contrast to similarity based aggregation strategies, the window generation methods provide a score to the proposal set to rank them and are considered faster [2,3,16,17].”
Luu, H. M., van Walsum, T., Son, M. H., Franklin, D. R., Nguyen, T. T. T., My, L. T., Moelker, A. D., Le, V. K., Vu, D. L., Le, N. H., et al. (2022) · Medical Image Analysis · doi:10.1016/j.media.2022.102422 · Cited by 8
Robustness Study of a Deep Convolutional Neural Network for Vehicle Detection in Aerial Imagery
Ilina, O. V., & Tereshonok, M. V. (2022) · Journal of Communications Technology and Electronics · doi:10.1134/s1064226922020048 · Cited by 7
Tunable Hybrid Proposal Networks for the Open World
Inkawhich, M. J., Inkawhich, N., Li, H. H., & Chen, Y. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00199 · Cited by 8
“Our core experimental methodology is to divide the COCO dataset [43]”
MethodGLNet for Target Detection in Millimeter Wave Images
Wang, S., Ye, Z., & Wang, Y. (2018) · International Conference on Multimedia and Image Processing (ICMIP) · doi:10.1145/3195588.3195609 · Cited by 7
“The traditional sliding window with different scales and steps may generate thousands of proposals per image with numerous false alarms [16].”
BackgroundNegin, F., & Brémond, F. (2019) · Sensors · doi:10.3390/s19194237 · Cited by 8
Singha, A., & Bhowmik, M. K. (2023) · Neural Computing and Applications · doi:10.1007/s00521-023-08390-7 · Cited by 6
Weakly Supervised Panoptic Segmentation for Defect-Based Grading of Fresh Produce
Knott, M., Odion, D., Sontakke, S., Karwa, A., & Defraeye, T. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00543 · Cited by 8
“For the average recall metric ( AR ), we follow the definition of Hosang et al. [10] evaluating a range of IOU thresholds.”
Depth-aware layered edge for object proposal
Liu, J., Ren, T., Bao, B., & Bei, J. (2016) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme.2016.7552901 · Cited by 6
“…window scoring based methods usually have higher efficiency, which are more suitable to serve as preprocessing for other applications, but they easily suffer the problem in proposal accuracy, i.e., they cannot provide acceptable proposals under high intersection over union (IoU) requirement [6].”
BackgroundMethodInfluentialDo we really need more training data for object localization
Li, H., Liu, Y., Zhang, X., An, Z., Wang, J., Chen, Y., & Tong, J. (2017) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2017.8296386 · Cited by 6
“AR summarizes the general proposal performance and is shown to correlate with the average precision (AP) performance of a detector better than other metrics [25].”
BackgroundRotated Mask Region-Based Convolutional Neural Network Detection for Parking Space Management System
Truong, L. N. H., Clay, E., Mora, O. E., Cheng, W., Singh, M., & Jia, X. (2022) · Transportation Research Record Journal of the Transportation Research Board · doi:10.1177/03611981221105066 · Cited by 6
“Like AP, average recall (AR) (86) is another popular numerical metric that can be used to compare detector performance.”
BackgroundVaishnavi, J., & Narmatha, V. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18886-6 · Cited by 7
“The above model depends more on unsupervised grouping which leads to diminishes repeatability [64].”
Gupta, N. D., Rajoo, R., & Jacob, P. J. (2023) · MALAYSIAN JOURNAL OF COMPUTING · doi:10.24191/mjoc.v8i1.20286 · Cited by 6
Rethinking Task and Metrics of Instance Segmentation on 3D Point Clouds
Arase, K., Mukuta, Y., & Harada, T. (2019) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw.2019.00505 · Cited by 6
“Following [21], we chose the proposal recall [9] as the evaluation metric and used PointNet as a feature extractor for a fair comparison.”
MethodA Universal Accelerated Coprocessor for Object Detection Based on RISC-V
Wu, D., Liu, Y., & Tao, C. (2023) · Electronics · doi:10.3390/electronics12030475 · Cited by 6
Gao, S., Xia, T., Hong, G., Zhu, Y., Chen, Z., Pan, E., & Xi, L. (2023) · Measurement · doi:10.1016/j.measurement.2023.113957 · Cited by 6
VSL-Net: Voxel structure learning for 3D object detection
Cao, F., Zhou, F., Tao, C., Xue, J., Gao, Z., Zhang, Z., & Zhu, Y. (2024) · Advanced Engineering Informatics · doi:10.1016/j.aei.2023.102348 · Cited by 7
Automatic Refinement Strategies for Manual Initialization of Object Trackers
Zhu, H., & Porikli, F. (2016) · IEEE Transactions on Image Processing · doi:10.1109/tip.2016.2633874 · Cited by 6
“tation for normalized-cut segmentation and an efficient algorithm for combinatorial merging of hierarchical regions [40], like [41] and [42].”
MethodFast region-based DPM object detection for autonomous vehicles
Ali, A., Olaleye, O. G., & Bayoumi, M. (2016) · IEEE International Midwest Symposium on Circuits and Systems (MWSCAS) · doi:10.1109/mwscas.2016.7870113 · Cited by 6
“For example, the state-of-the-art region proposal techniques [3] includes selective search, objectness, Binarized normed gradients (BING), constrained parametric min-cuts (CPMC) and Edge Boxes [12].”
BackgroundVisual tracking based on edge field with object proposal association
Kwon, J., & Lee, H. (2017) · Image and Vision Computing · doi:10.1016/j.imavis.2017.11.004 · Cited by 6
A generation method of insulator region proposals based on edge boxes
Zhao, Z., Zhang, L., Qi, Y., & Shi, Y. (2017) · Optoelectronics Letters · doi:10.1007/s11801-017-7201-8 · Cited by 6
Chen, W., Xing, J., Li, J., Yang, Q., & Zhou, Q. (2018) · International Conference on Digital Image Processing · doi:10.1117/12.2503030 · Cited by 6
Scale-Aware RPN for Vehicle Detection
Ding, L., Wang, Y., Laganière, R., Luo, X., & Fu, S. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-03801-4_43 · Cited by 6
Arapi, V., Santina, C. D., Bacciu, D., Bianchi, M., & Bicchi, A. (2018) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2018.00086 · Cited by 6
“Window proposal techniques (Hosang et al., 2016) have emerged mainly in object identification, where candidate bounding boxes are scored according to how likely they are to contain an object.”
BackgroundTewari, T., Sakhare, K. V., & Vyas, V. (2019) · Lecture notes in electrical engineering · doi:10.1007/978-981-13-7091-5_21 · Cited by 6
Airport Detection for Fixed-Wing Unmanned Aerial Vehicle Landing Using a Hierarchical Architecture
Hu, Y., Cao, Y., Ding, M., & Zhuang, L. (2019) · Journal of Aerospace Information Systems · doi:10.2514/1.i010615 · Cited by 6
Object Proposal Algorithms in the Wild: Are they Generalizable to Robot Perception?
Chan, D. M., & Riek, L. D. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8967830 · Cited by 6
“al [24] performed a study of object proposal algorithms on the PASCAL dataset and found that their repeatability is sensitive to some image perturbations such as rotation, shift, and image compression.”
BackgroundResultDeepFH segmentations for superpixel-based object proposal refinement
Wilms, C., & Frintrop, S. (2021) · Image and Vision Computing · doi:10.1016/j.imavis.2021.104263 · Cited by 8
Weakly Supervised Object BoundariesPreprint
Khoreva, A., Benenson, R., Omran, M., Hein, M., & Schiele, B. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1511.07803 · Cited by 6
“It is an enabling ingredient for many vision tasks such as image/video segmentation [1, 10], object proposals [15], object detection [35], and semantic labelling [2].”
BackgroundSelf-citationGooseDetectlion: A Fully Annotated Dataset for Lion-head Goose Detection in Smart Farms
Feng, Y., Wen, L., Guo, Y., Wang, Y., Tang, S., Yuan, Y., & Shen, L. (2024) · Scientific Data · doi:10.1038/s41597-024-03776-1 · Cited by 6
Learning proposal for front-vehicle detection
Tang, J., Li, J., & An, X. (2015) · Chinese Automation Congress · doi:10.1109/cac.2015.7382602 · Cited by 5
“In particular, given less than several hundreds proposals per image, our method can achieve over 93.3% recall at an overlap threshold of 0.7 per image within 1 second which is really challenging.”
MethodInfluentialHamroun, M., Lajmi, S., & Jallouli, M. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-20112-2 · Cited by 6
Text-Edge-Box: An Object Proposal Approach for Scene Texts Localization
Nguyen, D. C., Lu, S., Ouarti, N., & Mokhtari, M. (2017) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2017.149 · Cited by 5
“[41] shows that this criteria correlates well with the detection performance and it has been used to evaluate quality of the proposals in the EB [22] and the HyperNet [33].”
MethodA structure-guided approach to the prediction of natural image saliency
Liang, H., Jiang, M. X., Liang, R., & Zhao, Q. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.09.085 · Cited by 5
Ocampo, A. L. P. D., & Dadios, E. P. (2019) · International Journal of Advances in Intelligent Informatics · doi:10.26555/ijain.v5i3.426 · Cited by 7
Xu, Q., Wang, G., Li, Y., Shi, L., & Li, Y. (2021) · PLoS ONE · doi:10.1371/journal.pone.0251339 · Cited by 5
“Excellent classification accuracy and localization have been achieved by deep learning methods with high-quality region proposals [17, 18].”
BackgroundMask-Net: A Hardware-efficient Object Detection Network with Masked Region Proposals
Chen, H., & Hao, C. (2022) · IEEE International Conference on Application-specific Systems, Architectures and Processors (ASAP) · doi:10.1109/asap54787.2022.00030 · Cited by 6
“According to a recent survey [28], there are two ways to generate region proposals: grouping methods (such as Selective Search [29], [30]) and window scoring methods (such as Objectness [31], [32]).”
MethodDeep Learning-Based Object Detection Approach for Autonomous Vehicles
Albouchi, A., Hajjaji, M. A., & Mtibaa, A. (2022) · International Conference on Sciences and Techniques of Automatic Control and Computer Engineering · doi:10.1109/sta56120.2022.10018993 · Cited by 5
“First, region suggestions are extracted [3], followed by classification and correction of position coordinates.”
MethodAction detection in office scene based on deep convolutional neural networks
Yan, S.-Y., Yu-Di, A., Smith, J. S., & Zhang, B. (2016) · International Conference on Machine Learning and Cybernetics · doi:10.1109/icmlc.2016.7860906 · Cited by 5
“To evaluate the performance of open-source poselets person detector, we followed the basic approach in [16].”
MethodOptimizing a neural network for detection of moving vehicles in video
Bouma, H., Fischer, N. M., & Kruithof, M. C. (2017) · Counterterrorism, Crime Fighting, Forensics, and Surveillance Technologies · doi:10.1117/12.2277974 · Cited by 5
Deep Learning and Machine Learning for Object Detection in Remote Sensing Images
Yang, G., Luo, Q., Yang, Y., & Zhuang, Y. (2017) · Lecture notes in electrical engineering · doi:10.1007/978-981-10-7521-6_30 · Cited by 5
A Unified Object Detection Framework for Intelligent Retail Container Commodities
Zhao, L., Yao, J., Du, H., Zhao, J., & Zhang, R. (2019) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2019.8803536 · Cited by 5
“Firstly, region proposals [5, 6] are generated, and then they are classified and refined.”
BackgroundEffective object proposals: size prediction for pedestrian detection in surveillance videos
Qiu, J. W., Wang, L., Hu, Y. H., & Wang, Y. (2020) · Electronics Letters · doi:10.1049/el.2020.0850 · Cited by 5
“A comprehensive survey and comparison of object proposal generation methods can be found in [3].”
BackgroundMethodDolphin movement direction recognition using contour-skeleton information
Qi, H., Xue, M., Peng, X., Wang, C., & Jiang, Y. L. (2020) · Multimedia Tools and Applications · doi:10.1007/s11042-020-09659-y · Cited by 5
Indonesian food identification and detection in the smart nutrition box using faster-RCNN
Adinugroho, S., Adikara, P. P., Santoso, E., Amara, R., Septiana, K., & Anggita, K. D. (2020) · International Conference on Sustainable Information Engineering and Technology (SIET) · doi:10.1145/3427423.3427429 · Cited by 5
Superpixel-based Refinement for Object Proposal Generation
Wilms, C., & Frintrop, S. (2021) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr48806.2021.9412834 · Cited by 9
“We also compare to MCG [34], a top-performing non deep learning-based system [30].”
BackgroundMethodResultA region proposal algorithm using texture similarity and perceptual grouping
Taghizadeh, M., Chalechale, A., & Jannesari, A. (2021) · Journal of Ambient Intelligence and Humanized Computing · doi:10.1007/s12652-021-03296-5 · Cited by 5
Semantic image segmentation using region-based object detector
Shuhan, C., Wang, B., Li, J., & Xuelong, H. (2017) · IEEE International Conference on Electronic Measurement & Instruments · doi:10.1109/icemi.2017.8265872 · Cited by 5
“Experiments in [18] show that MCG (multiscale combinatorial grouping) [19] gets the best performance among all low-level feature based proposal generators.”
BackgroundHsu, C.-Y., & Li, W. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.04009 · Cited by 5
“Recent studies [18] have shown that the quality of proposals greatly affect the predictive performance of a network.”
InfluentialDeep Learning for Building Footprint Extraction Using UAV-Based Orthoimages
Ahmed, S. F., EL-Shazely, A. H., & Ahmed, W. (2024) · Journal of the Indian Society of Remote Sensing · doi:10.1007/s12524-024-02050-6 · Cited by 4
“The Region-based CNN (R-CNN) method for bound-ing-box object detection operates by narrowing the focus to a select set of potential object regions (Hosang et al., 2016).”
Towards Defect Detection of Transmission Line Insulator: A Dataset, Benchmarks and Challenges
Hu, Z., Zhai, Y., Zhao, Z., Wang, Q., Zhai, B., Yang, K., & Hu, P. (2025) · International Conference on Power and Renewable Energy (ICPRE) · doi:10.1109/icpre67300.2025.11274094 · Cited by 4
“The COCO evaluation metrics [25] primarily include Average Precision (AP) and Average Recall (AR).”
A robust multi-sensor fusion model against adversarial patch attack
El-Fatyany, A. (2026) · Wireless Networks · doi:10.1007/s11276-025-04065-4 · Cited by 5
“Furthermore, the performance of the proposed model is evaluated by employing Average Precision (AP) as the Area Under Precision-Recall(AUC) [ 79], where this metric is common for moderate cases that depend on the predicted and ground truth boxes.”
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.