The DOI matches its Crossref record; 56 of the 57 references it cites were found in a registry or an index.
Hide-and-Seek: Forcing a Network to be Meticulous for Weakly-Supervised Object and Action Localization
Open accessAbstract
We propose `Hide-and-Seek', a weakly-supervised framework that aims to improve object localization in images and action localization in videos. Most existing weakly-supervised methods localize only the most discriminative parts of an object rather than all relevant parts, which leads to suboptimal performance. Our key idea is to hide patches in a training image randomly, forcing the network to seek other relevant parts when the most discriminative part is hidden. Our approach only needs to modify the input image and can work with any network designed for object localization. During testing, we do not need to hide any patches. Our Hide-and-Seek approach obtains superior performance compared to previous methods for weakly-supervised object localization on the ILSVRC dataset. We also demonstrate that our framework can be easily extended to weakly-supervised action localization.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
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?
In the citing paper
“However, various types of industrial image defects do not meet all the above assumptions [42,43].”
From An Improved VGG19 Transfer Learning Strip Steel Surface Defect Recognition Deep Neural Network Based on Few Samples and Imbalanced Datasets (2021) · cites this paper as [43]
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
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Going deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380
Fully convolutional networks for semantic segmentation
Long, J., Shelhamer, E., & Darrell, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298965 · Cited by 38,834
(2014) · Cited by 33,419
Rich 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
(2016) · Cited by 31,590
Girshick, R. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.169 · Cited by 28,105
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S., & Szegedy, C. (2015) · arXiv · doi:10.48550/arxiv.1502.03167 · Cited by 27,818
Visualizing and understanding convolutional networks
Zeiler, M. D., & Fergus, R. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10590-1_53 · Cited by 15,839
Flagged first, then by citations
1 of the 57 references couldn’t be matched to a known paper.
Research works printed but not linked: 5
- jiang, 2014, THUMOS Challenge Action Recognition with A Large Number of Classes
- jhuang, 2013, Towards undestanding action recognition, ICCV
- 10.1109/CVPR.2016.311 · DOI · Paper page
- cinbis, 2015, Weakly supervised object localization with multi-fold multiple instance learning
- wan, 2013, Regularization of neural network using dropconnect, ICML
Cited by
Sorted by citations
Items 501–600 of 720
Background-Aware Robust Context Learning for Weakly-Supervised Temporal Action Localization
Kim, J., & Cho, J. (2022) · IEEE Access · doi:10.1109/access.2022.3183789 · Cited by 4
“Hideand-seek [22] and MAAN [27] introduced methods that suppress the dominant response by erasing the most discriminative parts.”
One Weird Trick to Improve Your Semi-Weakly Supervised Semantic Segmentation Model
Bae, W., Noh, J., Asadabadi, M. J., & Sutherland, D. J. (2022) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2022/389 · Cited by 4
Aggregation of attention and erasing for weakly supervised object localization
Koo, B., Choi, H.-S., & Kang, M. (2022) · Image and Vision Computing · doi:10.1016/j.imavis.2022.104598 · Cited by 4
Negative Contrast: A Simple and Efficient Image Augmentation Method in Crop Disease Classification
Li, J., Yin, Z., Li, D., & Zhao, Y. (2023) · Agriculture · doi:10.3390/agriculture13071461 · Cited by 4
Exploring Zero-Shot Semantic Segmentation with No Supervision Leakage
Wang, Y., & Tian, Y. (2023) · Electronics · doi:10.3390/electronics12163452 · Cited by 5
Few-Shot Common Action Localization via Cross-Attentional Fusion of Context and Temporal Dynamics
Lee, J., Jain, M., & Yun, S. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00937 · Cited by 4
“Temporal action localization [2, 4, 33, 23, 31, 18, 15, 16, 20] have been widely studied in fully or weakly-supervised manner.”
BackgroundWu, C., Li, M., Gao, Y., Xie, X., Ng, W. W. Y., & Musyafa, A. (2024) · Machine Intelligence Research · doi:10.1007/s11633-023-1455-3 · Cited by 4
Improving Weakly Supervised Object Localization by Uncertainty Estimation of pseudo supervision
Chen, X., Ma, A. J., Guo, N., & Chen, J. (2021) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme51207.2021.9428113 · Cited by 4
Diagnostics of Continuous Dynamic Systems Using Logic Functions
Shalobanov, S. V., & Shalobanov, S. S. (2021) · Dynamics of Systems Mechanisms and Machines · doi:10.1109/dynamics52735.2021.9653718 · Cited by 4
Dual-attention Focused Module for Weakly Supervised Object LocalizationPreprint
Zhou, Y., Chen, Z., Shen, H., Liu, Q., Zhao, R., & Liang, Y. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.04813 · Cited by 4
“To solve this problem, Hide-and-Seek method is proposed to hide patches in a training image randomly, forcing the network to seek other relevant parts (Singh and Lee, 2017).”
MethodInfluentialTowards Train-Test Consistency for Semi-supervised Temporal Action LocalizationPreprint
Lin, X., Shou, Z., & Chang, S. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.11285 · Cited by 5
“It is only constrained by the video-level classification loss, which makes the detection model only capable of detecting the most discriminative part of an action [43].”
BackgroundMethodInfluentialGlobal context-aware attention model for weakly-supervised temporal action localization
Fu, W., Zhang, W., Long, J., Srivastava, G., & Liu, S. (2025) · Alexandria Engineering Journal · doi:10.1016/j.aej.2025.05.006 · Cited by 3
Liu, Z. L., Zhang, Z., & Li, S. (2026) · Agriculture · doi:10.3390/agriculture16080875 · Cited by 3
Learning Deep Asymmetric Tolerant Part Representation
Yu, X., Zhao, Y., Gao, Y., & Xiong, S. (2022) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2022.3222644 · Cited by 3
POTLoc: Pseudo-label Oriented Transformer for point-supervised temporal Action Localization
Vahdani, E., & Tian, Y. (2024) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2024.104044 · Cited by 3
Adaptive Prototype Learning for Weakly-Supervised Temporal Action Localization
Luo, W., Ren, H., Zhang, T., Yang, W., & Zhang, Y. (2024) · IEEE Transactions on Image Processing · doi:10.1109/tip.2024.3431915 · Cited by 4
“The first group of works adopts the erasing mechanism [39], [40], [41] or a multi-branch architecture [16] to improve the localization completeness.”
Breakthrough in fine-grained video anomaly detection on highway: New benchmark and model
Meng, C., Wang, X., Zhang, C., Mao, Z., Shen, J., & Cheng, Z. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.129127 · Cited by 3
How to Save your Annotation Cost for Panoptic Segmentation?
Du, X., Jiang, C., Xu, H., Zhang, G., & Li, Z. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i2.16216 · Cited by 3
“Weakly/Semi-supervised approaches for semantic segmentation (Wang et al. 2018; Wei et al. 2017; Singh and Lee 2017; Hung et al. 2018; Ouali, Hudelot, and Tami 2020; Papandreou et al. 2015) and instance segmentation (Hu et al.”
BackgroundDuan, W., Chiang, Y., Leyk, S., Uhl, J. H., & Knoblock, C. A. (2021) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata52589.2021.9671875 · Cited by 3
“Weakly supervised object detection [19], [20], [21], [22], [23], [12], [11], [24], [25], [26], [27], aiming to localize objects with image-level annotations, has attracted attention because bounding box annotations require intensive manual work.”
BackgroundGrid self-occlusion: a grid self-occlusion data augmentation for better classification
Deng, X. F., Zhao, H., Zhang, H., Zhang, J., & Mei, B. (2022) · Signal Image and Video Processing · doi:10.1007/s11760-022-02278-0 · Cited by 3
“The experiment mainly compares with RE, cutout, HAS, grid mask data augmentation methods, the results are shown in Table 5 It can be seen that our method improves by 4.19%.”
BackgroundInfluentialComplementary Attention Network for Weakly Supervised Temporal Action Localization
Dou, P., & Hu, H. (2023) · Neural Processing Letters · doi:10.1007/s11063-023-11156-w · Cited by 3
Point-supervised temporal action localisation based on multi-branch attention
Liu, S., Zhang, Y., & Srivastava, G. (2023) · Enterprise Information Systems · doi:10.1080/17517575.2023.2197318 · Cited by 3
“Currently, the tasks of weakly supervised temporal action localisation are mainly divided into three methods: the methods based on the erasure (Singh, and Jae Lee 2017; Wang et al. 2017; Zhang et al. 2019), the methods based on attention mechanism (Lee, Youngjung, and Byun 2020; Luo et al. 2021;…”
MethodPositive-weighting feature enhancement for weakly supervised object localization
Shieh, J.-L., Yu, S., & Ruan, S.-J. (2023) · Pattern Recognition Letters · doi:10.1016/j.patrec.2023.04.017 · Cited by 3
Dilation-erosion for single-frame supervised temporal action localization
Wang, B., Song, Y., Wang, F., Zhao, Y., Shu, X., & Yan, R. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-15196-1 · Cited by 3
Improving Data-Efficient Fossil Segmentation via Model Editing
Panigrahi, I., Manzuk, R. A., Maloof, A. C., & Fong, R. (2023) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw59228.2023.00511 · Cited by 3
“Others partially occlude images during training as a data augmentation technique to improve model robustness [5, 7, 33] and/or localization performance [38].”
MethodData Augmentation Using Corner CutMix and an Auxiliary Self-Supervised Loss
Fang, F., Hoang, N. M., Xu, Q., & Lim, J. (2023) · Proceedings - International Conference on Image Processing · doi:10.1109/icip49359.2023.10222009 · Cited by 3
“To alleviate over-fitting caused by a CNN focusing too much on a small set of activation or small set region on input, effective regularization techniques, Image Mix [4, 5, 6, 7, 8] and random region erasing [9, 10, 11, 12, 13] have been proposed.”
MethodVisibility Estimation Based on Weakly Supervised Learning under Discrete Label Distribution
Yan, Q. Y., Sun, T., Zhang, J., & Xun, L. (2023) · Sensors · doi:10.3390/s23239390 · Cited by 3
Ma, S., Hou, B., Li, Z., Wu, Z., Guo, X., Yang, C., & Jiao, L. (2023) · Neurocomputing · doi:10.1016/j.neucom.2023.127213 · Cited by 3
Unsupervised Object Localization with Representer Point Selection
Song, Y., Jang, S.-W., Katabi, D., & Son, J. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00601 · Cited by 3
“In this section, we discuss how we leverage the represen-ter point selection [47] to solve unsupervised object localization (UOL) task, and our extension to weakly supervised object localization (WSOL) and zero-shot transferring.”
Boosting Weakly Supervised Object Localization and Segmentation With Domain Adaption
Zhu, L., She, Q., Chen, Q., Ren, Q., & Lu, Y. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3411036 · Cited by 3
“For activating more object parts, different technics have been adopted by designing data augmentation strategies [12], [18], [19], [20], modifying network structures [12], [14], [15], [21], adding additional training streams/stages [13], [22], [23], [24], and proposing post refining processes [16],…”
BackgroundMethodInfluentialWeakly Supervised Object Localization With Progressive Activation Diffusion
Xu, C., Hui, L., Xie, J., & Yang, J. J. (2025) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2025.3540898 · Cited by 3
“For instance, HAS [18] randomly hides image patches in the training stage to search for other relevant parts of objects.”
Localizing Common Objects Using Common Component Activation Map
Li, W., Jafari, O. H., & Rother, C. (2019) · Cited by 3
Hierarchical class grouping with orthogonal constraint for class activation map generation
Meng, F., Huang, K., Li, H., Chen, S., Wu, Q., & Ngan, K. N. (2020) · Neural Computing and Applications · doi:10.1007/s00521-020-05416-2 · Cited by 3
“It is an important task in computer vision as it can locate regions of a specific class from weak labels and thus can help many weakly supervised tasks such as segmentation [2, 7, 10, 12, 14, 25], detection [21, 24, 27, 29] and recognition [3, 5, 9, 20, 22].”
BackgroundConvolutional STN for Weakly Supervised Object Localization
Meethal, A., Pedersoli, M., Belharbi, S., & Granger, É. (2021) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr48806.2021.9412029 · Cited by 4
“Recently, several weakly supervised learning (WSL) techniques have been proposed for object localization, to alleviate the need for such expensive fine-grained annotations [12]–[14].”
InfluentialHierarchical complementary learning for weakly supervised object localization
Benassou, S. N., Shi, W., Jiang, F., & Benzine, A. (2021) · Signal Processing Image Communication · doi:10.1016/j.image.2021.116520 · Cited by 3
A Weakly Supervised Defect Detection Based on Dual Path Networks and GMA-CAM
Xie, H., & Lin, S. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-87355-4_39 · Cited by 4
Video segmentation via target objectness constraint and multi-head soft aggregation
Wang, H., Liu, W., Xing, W., & Zhang, S. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.08.065 · Cited by 3
CyclicShift: A Data Augmentation Method For Enriching Data Patterns
Lu, H. M., Cheng, X., Xia, W., Pan, D., Liu, M., Xie, T., Wang, X., & Liu, M. (2022) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3503161.3548188 · Cited by 3
“Other data augmentation methods based on image information deletion also have excellent performance, such as Cutout [5], Hideand-Seek [28], Random erasing [37], GridMask [3], etc.”
BackgroundMethodKoo, B., Choi, H.-S., & Kang, M. (2021) · Multidimensional Systems and Signal Processing · doi:10.1007/s11045-021-00778-9 · Cited by 3
“HaS (Hide-and-seek) (Singh & Lee 2017) and CutMix (Yun et al. 2019) are data augmentation approaches to WSOL.”
BackgroundMethodComplementary Temporal Classification Activation Maps in Temporal Action Localization
Wang, L., Zhu, S., Li, Z., & Fang, Z. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-88007-1_31 · Cited by 3
HistoSegResT: A Weakly Supervised Learning Method for Histopathology Image Segmentation
Gu, W., Wang, S., Zhao, S., Wan, L., & Zhu, Z. (2022) · International Conference on Image and Graphics Processing (ICIGP) · doi:10.1145/3512388.3512416 · Cited by 3
“Existing bottom-up WSL methods could be categorized as follows: (1) spatial pooling techniques over activation maps [19, 20], (2) multi-instance learning [21] and (3) attend-and-erase based methods [22–24].”
MethodAutoLoc: Weakly-supervised Temporal Action LocalizationPreprint
Shou, Z., Gao, H., Zhang, L., Miyazawa, K., & Chang, S.-F. (2018) · European Conference on Computer Vision · doi:10.48550/arxiv.1807.08333 · Cited by 290
“Recently, Singh et al. designed Hide-and-Seek [56] to address the challenge that weakly-supervised detection methods usually focus on the most discriminative parts while neglect other relevant parts of the target instance.”
BackgroundInfluentialWeakly-Supervised Learning for Tool Localization in Laparoscopic VideosPreprint
Vardazaryan, A., Mutter, D., Marescaux, J., & Padoy, N. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1806.05573 · Cited by 3
“Some recent works have used this fact to successfully localize objects in images without explicitly training for localization [12,15,17].”
Weakly Supervised Temporal Action Localization with Segment-Level LabelsPreprint
Ding, X., Wang, N., Gao, X., Li, J., Wang, X., & Liu, T. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.01598 · Cited by 3
“Given only category labels, most of methods [1,9,14,10,8,17] tend to generate class activation sequences (CAS) from a classification loss.”
BackgroundInfluentialLID 2020: The Learning from Imperfect Data Challenge ResultsPreprint
Wei, Y., Zheng, S., Cheng, M., Zhao, H., Wang, L., Ding, E., Yang, Y., Torralba, A. B., Liu, T., Sun, G., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.11724 · Cited by 4
“…[1, 2, 6, 7, 13–17, 20, 22–25, 27, 28, 30, 32, 35– 38, 40–42, 44–46, 48–57, 59–61, 63, 66, 67, 74], weakly supervised scene parsing using point supervision [39] and weakly supervised object localization [3–5, 9–11, 18, 19, 21, 26, 29, 31, 33, 34, 43, 47, 58, 62, 64, 65, 68–72, 75], respectively.”
Self-Supervised Learning for Fine-Grained Visual CategorizationPreprint
Maaz, M., Rasheed, H., & Gaddam, D. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.08788 · Cited by 3
“More related to this work, [32] proposed a weakly-supervised hide and seek framework to localize only the most discriminative parts of an object rather than all relevant parts.”
BackgroundMONCAE: Multi-Objective Neuroevolution of Convolutional AutoencodersPreprint
Dimanov, D., Balaguer‐Ballester, E., Singleton, C., & Rostami, S. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.11914 · Cited by 7
“There is a plethora of different attempts for solving this problem, focusing both in the ’data space’ (Wang et al., 2018; Singh & Lee, 2017) as well as in the ’algorithmic space’ (Hinton et al., 2015; Frankle & Carbin, 2018).”
BackgroundMaskBlur: Spatial and Angular Data Augmentation for Light Field Image Super-Resolution
Chao, W., Duan, F., Guo, Y., & Wang, G. (2024) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2024.3521781 · Cited by 3
“Traditional Methods: Traditional methods for LF image SR relied on the geometric structure [26] and mathematical modeling [27] of the LF, which utilized projection and optimization techniques to super-resolve LR images.”
Snippet-Inter Difference Attention Network for Weakly-Supervised Temporal Action Localization
Zhou, W., Lin, K., Hu, W., Xie, C., Su, T., Hu, H., & Tan, Y. (2025) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2025.3535336 · Cited by 2
Self-supervised temporal adaptive learning for weakly-supervised temporal action localization
Sheng, J., Yu, J., Li, Z., Li, A., & Ge, Y. (2025) · Information Sciences · doi:10.1016/j.ins.2025.121986 · Cited by 2
Liao, G., Jogan, M., Koushik, S., Eaton, E., & Hashimoto, D. A. (2025) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61041.2025.00778 · Cited by 2
“Training strategies such as randomly masking out images [23] or feature maps [12,26] can refine the CAM by enhancing activation of novel features.”
Cross-Task Relation-Aware Consistency for Weakly Supervised Temporal Action Detection
Yang, W.-F., Ren, H., Zhang, T., Zhang, Z., Zhang, Y., & Wu, F. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3594178 · Cited by 3
“In Hide-and-Seek [49], a random erasing mechanism is used to randomly zero out input segments during training.”
Abdelhalim, I., Almalki, Y. E., Abdallah, A., Karam, R., Alduraibi, S. K., Basha, M. A. A., Mohamed, H., Ghazal, M., Mahmoud, A., Alghamdi, N. S., et al. (2025) · Medical Physics · doi:10.1002/mp.18119 · Cited by 2
Wu, Z., Zheng, S., Xu, Y., Wang, L., Wei, Z., & Hua, G. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3646464 · Cited by 2
“Singh et al. [23] randomly hide patches of an object to force the network seek other relevant regions.”
Kasem, M. S., Mahmoud, M., Senussi, M. F., Abdalla, M. I., & Kang, H.-S. (2026) · Mathematics · doi:10.3390/math14010187 · Cited by 2
Shallow Feature-driven Dual-edges Localization Network for Weakly Supervised Localization
Hui, W., Gu, G., & Wang, B. (2023) · Machine Intelligence Research · doi:10.1007/s11633-022-1368-6 · Cited by 2
Kim, D. M., Cha, S., & Kang, B. (2023) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip59821.2023.10402750 · Cited by 2
“Backbone Top-1 Top-5 GT-known CAM [2] 43.35 54.44 58.97 HaS [14] MobileNetV1 42.73 - 60.12 ADL [15] [16] 43.01 - - RCAM [17] 44.78 - 61.69 FAM [18] 46.24 - 62.05 CAM [2] 46.29 58.19 62.68 SPG [19] 48.60 60.00 64.69 DANet [7] 47.53 58.28 - PSOL [5] 54.82 63.25 65.21 I2C [20] InceptionV3 53.11 64.13…”
MethodInfluentialWeakly supervised temporal action localization: a survey
Li, R., Zhang, T., & Zhang, R. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18554-9 · Cited by 2
“Toresolvethiscontradictionbetweenclassification and localization in the image domain, the Hide-and-Seek framework [27] is proposed.”
Wei, R., Zhang, W., Kou, W., Rey, C., Marshall, H. R., & Chiu, B. C. Y. (2024) · SPIE Medical Imaging: Image Processing · doi:10.1117/12.3006245 · Cited by 2
“5,9,10 These approaches cover only small and the most discriminative regions, which may be in the background and irrelevant to the foreground being classified.”
BackgroundImproving weakly-supervised object localization using adversarial erasing and pseudo label
Kang, B., Cha, S., & Lee, Y. (2024) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2024.108456 · Cited by 3
“HaS (Singh and Lee, 2017) - Random erasing applied to images - Adversarial erasing applied to feature maps ACoL (Zhang et al., 2018a) -”
BackgroundMethodInfluentialSelf-citationGCLNet: Generalized Contrastive Learning for Weakly Supervised Temporal Action Localization
Wang, J., Kong, D., & Yin, B. (2025) · IEEE Transactions on Big Data · doi:10.1109/tbdata.2025.3528727 · Cited by 2
“The first type is the erasure method [11], [17], [18], [19], which designed an anti-erasure strategy by iteratively erasing the most discriminative region to mine the sub-region with ambiguous features, and then learning the integrity of the action.”
SSMask: Salient Super-Pixel Mask Data Augmentation for Object Detection
Zhao, H., Huang, J., Deng, H., & Deng, X. F. (2024) · International Conference on Pattern Recognition and Artificial Intelligence (PRAI) · doi:10.1109/prai62207.2024.10827066 · Cited by 2
“Such as Random-Erasing[13], Cut-Out[14], Grid-Mask[15], and Hide-and-Seek[16].”
Hide-CAM: Finding Multiple Discriminative Regions in Weakly Supervised Location
Xu, J., Sheng, S., Wei, H., & Guo, J. (2019) · IEEE Access · doi:10.1109/access.2019.2939267 · Cited by 2
“[13] proposed ‘‘Hide-and-Seek’’ by randomly hiding the patches of input images to force the network to look for other discriminative parts.”
BackgroundWeakly Supervised Learning for Object Localization Based on an Attention Mechanism
Park, N., & Ko, H. (2021) · Applied Sciences · doi:10.3390/app112210953 · Cited by 2
Multiple Instance Learning through Explanation by Using a Histopathology Example
Konstantinov, A. V., & Utkin, L. V. (2022) · Conference of Open Innovations Association (FRUCT) · doi:10.23919/fruct54823.2022.9770901 · Cited by 2
“The procedure of the image fill is similar to the algorithm “Hide-and-Seek” [27], but not the same.”
BackgroundMethodInfluentialExtracting Action Sensitive Features to Facilitate Weakly-Supervised Action Localization
Kang, Z., Wang, L., Liu, Z., Zhang, Q., & Zheng, N. (2019) · IFIP advances in information and communication technology · doi:10.1007/978-3-030-19823-7_15 · Cited by 3
“Because labeling action boundary in untrimmed video is expensive, some researchers [25,26,30,33,38] proposed to use video-level action annotation to produce snippet level action localization results, which greatly reduced demand for human laboring and yield comparable performance.”
BackgroundA Text Localization Method Based on Weak Supervision
Zhang, J., Du, C., Feng, Z., Wang, Y., & Wang, C. (2019) · IAPR International Conference on Document Analysis and Recognition (ICDAR) · doi:10.1109/icdar.2019.00129 · Cited by 2
“[31] use a strategy to hide the patches of input images randomly so that it can find other discriminative parts of the object.”
MethodAction Recognition in Real-World Videos
Sultani, W., Arshad, Q. A., & Chen, C. (2020) · Springer eBooks · doi:10.1007/978-3-030-03243-2_846-1 · Cited by 2
Luo, L., Yuan, C., Zhang, K., Jiang, Y., Zhang, Y., & Zhang, H. (2020) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme46284.2020.9102801 · Cited by 2
“Another kind, model agnostic methods [5, 6, 7], focuses on finding the area that will make the most drastic influence Fig.”
MethodSemantic feature augmentation for fine-grained visual categorization with few-sample training
Guan, X., Yang, Y., Wang, Z., & Li, J. (2021) · ACM International Conference on Multimedia in Asia (MMAsia) · doi:10.1145/3444685.3446264 · Cited by 2
“Before our work, random spatial image augmentation methods [17, 22], such as image cropping and image dropping, have been proposed and proved to be effective in improving the robustness of deep models.”
MethodDiagnostic Models of Sensitivity of Automatic Control Systems
Shalobanov, S. V., & Shalobanov, S. S. (2022) · Dynamics of Systems Mechanisms and Machines · doi:10.1109/dynamics56256.2022.10014824 · Cited by 2
Class Specialized Knowledge Distillation
Wang, L., Rhodes, A. D., & Feng, W. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-26284-5_24 · Cited by 3
On the Study of Data Augmentation for Visual Place Recognition
Jang, S., & Kim, U.-H. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3301778 · Cited by 2
“Second, image erasing techniques such as Cutout [28], Hide and Seek [29], Random Erasing [30], GridMask [31] help to handle clutter and occlusion due to non-informational elements.”
MethodAction Recognition From Single Timestamp Supervision in Untrimmed Videos
Moltisanti, D., Fidler, S., & Damen, D. (2019) · Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2019.01015 · Cited by 71
“When only a few different actions are present in an untrimmed video, video-level supervision can prove sufficient to learn the actions even from long videos, as recently shown in [32, 22, 31, 28, 23].”
BackgroundWeakly-Supervised Action Recognition and Localization via Knowledge Transfer
Shi, H., Zhang, X., & Li, C. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-31654-9_18 · Cited by 2
Regional Patch-Based Feature Interpolation Method for Effective Regularization
Jang, S., Jin, K., An, J., & Kim, Y. (2020) · IEEE Access · doi:10.1109/access.2020.2974328 · Cited by 2
“Besides, for some tasks such as object detection, a dropout with various techniques including attention mechanism have been examined [29], [30].”
Wang, Q., Song, Y., Zou, R., & Shu, X. (2022) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2022.103590 · Cited by 2
A Robust System to Detect and Explain Public Mask Wearing Behavior
Gupta, A., & Srivastava, B. (2022) · Studies in computational intelligence · doi:10.1007/978-3-031-14771-5_11 · Cited by 2
Multi-scale discriminative Region Discovery for Weakly-Supervised Object LocalizationPreprint
Lv, P., Yu, H., Xue, J., Cheng, J., Cui, L., Zhou, B., Xu, M., & Yang, Y. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.10698 · Cited by 2
“Different from Grad-CAM and Grad-CAM++ built on existing classification network without extra modification, other researchers devised different networks or structures [20], [21], [22], [23] to solve the mentioned problems above.”
BackgroundMethodInfluentialSelf-supervised Robust Object Detectors from Partially Labelled DatasetsPreprint
Abbasi, M., Laurendeau, D., & Gagné, C. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.11549 · Cited by 2
“Inspired by [18], we make use of patch-drop at test time of h in order to estimate the true class of given a RoI more accurately.”
Improving Augmentation and Evaluation Schemes for Semantic Image SynthesisPreprint
Katiyar, P., & Khoreva, A. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.12636 · Cited by 2
“Inspired by the task-specific augmentation studies in other vision applications [15, 14, 51, 3, 60], in this work, we propose an augmentation method specifically designed to overcome the above mentioned limitations of the SIS models.”
MethodBroomé, S., Ask, K., Rashid, M., Andersen, P. H., & Kjellström, H. (2021) · arXiv (Cornell University) · Cited by 2
Semantically Proportional Patchmix for Few-Shot Learning
Wang, J., Xu, J., Pan, Y. R., & Xu, Z. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747083 · Cited by 1
Learning Consistency From High-Confidence Pseudo-Labels for Weakly Supervised Object Localization
Sun, K., & Zhu, J. (2023) · IEEE Access · doi:10.1109/access.2023.3246259 · Cited by 2
Enhancing Noisy Label Facial Expression Recognition With Split and Merge Consistency Regularization
Kim, J., Kwon, J., Kim, M., Lee, E., & Kim, Y. (2023) · IEEE Access · doi:10.1109/access.2023.3339763 · Cited by 1
“Inspired by previous research [22], [23], [24] that extracted high-level CAMs by manipulating and transforming images, we propose an approach that splits a facial image into two and merges the images.”
MethodCut-Stitch: A Simple and Effective Data Augmentation Method for Industrial Inspection
Hu, H., Hong, J., Song, K., Zhu, J., & Ren, W. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-70359-1_24 · Cited by 1
Concept Visualization: Explaining the CLIP Multi-modal Embedding Using WordNet
Giulivi, L., & Boracchi, G. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10650232 · Cited by 1
“Restrictions apply.”
Liu, T., Lam, K., & Bao, B.-K. (2024) · International Journal of Computer Vision · doi:10.1007/s11263-024-02279-1 · Cited by 2
“(2) The erasing-based (Singh and Lee, 2017; Zhang et al., 2019; Zhong et al., 2018), multi-attention-based (Huang et al., 2021; Liu et al., 2021; Islam et al., 2021) and multi-branch-based (Liu et al., 2019; Huang et al., 2022) methods aim to obtain complete localization results, by encouraging the…”
Li, B., Shi, E., Ruan, H., Jiang, Z., Li, X., Wang, K., & Wang, S. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.126273 · Cited by 2
A Weakly Supervised Semantic Segmentation Method Based on Improved Conformer
Shen, X., & Wang, M. (2025) · Computers, materials & continua/Computers, materials & continua (Print) · doi:10.32604/cmc.2025.059149 · Cited by 1
“Kumar et al. [20] randomly hid patches in the image during training as a way to motivate the network to explore other relevant parts.”
Fujimori, I., Oono, M., & Shishibori, M. (2025) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2025.104404 · Cited by 1
Li, B., Chen, Q., Lv, B., Wang, K., & Li, X. (2025) · Journal of Electronic Imaging · doi:10.1117/1.jei.34.1.013056 · Cited by 1
“Hide-and-seek (HAS) 11 employs a random erasure strategy forcing the classification network to focus on relevant parts of the target.”
InfluentialWeakly supervised semantic segmentation based on affinity vector consistency
Bi, X., Xu, P., Fan, J., Liu, B., Ma, J., & Xiao, B. (2025) · Scientia Sinica Informationis · doi:10.1360/ssi-2024-0222 · Cited by 1
Small object matters in weakly supervised object localization
Hwang, D., Oh, S. J., & Choe, J. (2025) · Neurocomputing · doi:10.1016/j.neucom.2025.130494 · Cited by 1
IPF-RDA: An Information-Preserving Framework for Robust Data Augmentation
Yang, S., Yang, H., Guo, S., Shen, F., & Zhao, J. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3613005 · Cited by 1
“…significantly improve a number of state-of-the-art (SOTA) and widely adopted DA methods, including but not limited to Cutout [4], AdvMask [10], GridMask [19], Hide-and-Seek (HaS) [20], CutMix [5], RandomErasing [21], AutoAugment [6], Fast-AutoAugment [22], RandAugment [7], TrivialAugment [8], etc.”
Action-to-Action Diffusion Network for Weakly Supervised Temporal Action Localization
Zou, Y., Zhao, Q., Sarker, P. K., Yang, L., & Wang, B. (2025) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2025.3613174 · Cited by 1
“In contrast, the Hide-and-Seek mask [43] performs similarly to the random masking strategy by randomly hiding a different set of parts at each training iteration.”
Weakly Supervised Temporal Action Localization Based on Feature Enhancement
Zhang, H., & Yao, Y. (2026) · Cognitive Computation · doi:10.1007/s12559-026-10560-3 · Cited by 1
“For instance, the Hide-and-Seek[22] model randomly discards video segments during training, compelling the model to attend to other relevant parts and thus capture the full scope of the action.”
Weak Supervised Surface Defect Detection Method Based on Selective Search and CAM
Xiao, Y., Song, Y., & Tang, X. (2019) · Chinese Automation Congress · doi:10.1109/cac48633.2019.8996269 · Cited by 1
Learning a Weight Map for Weakly-Supervised Localization
Shaharabany, T., & Wolf, L. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10094940 · Cited by 1
“For example, [35] covers the discriminative parts during the training of the classifier in order to have it rely on additional regions.”
Weakly-Supervised Action Detection Guided by Audio Narration
Ye, K., & Kovashka, A. (2022) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw56347.2022.00159 · Cited by 1
“As for the weakly supervised action detection models [27, 39, 40, 42, 44, 53], their data in most cases only involves a single action per video.”
BackgroundMethod
Metadata
- Authors
- Singh, K. K., & Lee, Y. J.
- Affiliations
- Intel (United States) (US)
- University of California, Davis (US)
- Venue
- 2017 IEEE International Conference on Computer Vision (ICCV)
- Publisher
- IEEE
- Published
- 2017-10
- Pages
- 3544-3553
- OpenAlex
- W2606520630
- Cited by
- 724
- References
- 57
- Citations by year
- 2026: 19 · 2025: 49 · 2024: 74 · 2023: 121 · 2022: 129 · 2021: 137 · 2020: 98 · 2019: 70 · 2018: 25 · 2017: 2
- FWCI
- 15.59
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Nvidia
- Amazon Web Services
- 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
- weakly supervised object localization, data augmentation, image classification, video action recognition
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
