Abstract
Temporal Action Localization (TAL) is an important task of various computer vision topics such as video understanding, summarization, and analysis. In the real world, the videos are long untrimmed and contain multiple actions, where the temporal boundaries annotations are required in the fully-supervised learning setting for classification and localization tasks. Since the annotation task is costly and time-consuming, the trend is moving toward the weakly-supervised setting, which depends on the video-level labels only without any additional information, and this approach is called weakly-supervised Temporal Action Localization (WTAL). In this survey, we review the concepts, strategies, and techniques related to the WTAL in order to clarify all aspects of the problem and review the state-of-the-art frameworks of WTAL according to their challenges. Furthermore, a comparison of models’ performance and results based on benchmark datasets is presented. Finally, we summarize the future works to allow the researchers to improve the model's performance.
Similar content being viewed by others
References
Lin X, Shou Z, Chang S-F (2019) Towards train-test consistency for semi-supervised temporal action localization, [Online]. Available: http://arxiv.org/abs/1910.11285
Ma F et al (2020) SF-Net: single-frame supervision for temporal action localization, [Online]. Available: http://arxiv.org/abs/2003.06845
Ding X, Wang N, Gao X, Li J, Wang X, and Liu T (2020) Weakly supervised temporal action localization with segment-level labels, 1(c), [Online]. Available: http://arxiv.org/abs/2007.01598
Sun C, Shetty S, Sukthankar R, and Nevatia R (2015) Temporal localization of fine-grained actions in videos by domain transfer from web images. In: MM 2015 - Proc. 2015 ACM Multimed. Conf., pp. 371–380. https://doi.org/10.1145/2733373.2806226
Park J, Lee J, Jeon S, Kim S, and Sohn K (2019) Graph regularization network with semantic affinity for weakly-supervised temporal action localization. In: Proceedings - international conference on image processing, ICIP, 2019:3701–3705. https://doi.org/10.1109/ICIP.2019.8803589
Nguyen P, Han B, Liu T, and Prasad G (2018) Weakly supervised action localization by sparse temporal pooling network. In: Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit, pp. 6752–6761. https://doi.org/10.1109/CVPR.2018.00706
Narayan S, Cholakkal H, Khan FS, and Shao L (2019) 3C-Net: category count and center loss for weakly-supervised action localization. Proc IEEE Int Conf Comput Vis 2019: 8678–8686. https://doi.org/10.1109/ICCV.2019.00877
Wang C, Cai H, Zou Y, and Xiong Y (2021) RGB stream is enough for temporal action detection, [Online]. Available: http://arxiv.org/abs/2107.04362
Alwassel H, Giancola S, and Ghanem B (2020) TSP: temporally-sensitive pretraining of video encoders for localization tasks, [Online]. Available: http://arxiv.org/abs/2011.11479
Nawhal M and Mori G (2021) Activity graph transformer for temporal action localization, [Online]. Available: http://arxiv.org/abs/2101.08540
Alwassel H, Pardo A, Heilbron FC, Thabet A, and Ghanem B (2019) RefineLoc: iterative refinement for weakly-supervised action localization, [Online]. Available: http://arxiv.org/abs/1904.00227
Bojanowski P et al (2014) Weakly supervised action labeling in videos under ordering constraints. Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), vol. 8693 LNCS, no. Part 5, pp. 628–643. https://doi.org/10.1007/978-3-319-10602-1_41
Huang DA, Fei-Fei L, and Niebles JC (2016) Connectionist temporal modeling for weakly supervised action labelling. Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), vol. 9908 LNCS, pp. 137–153. https://doi.org/10.1007/978-3-319-46493-0_9
Yang H, He X, Porikli F (2018) One-shot action localization by learning sequence matching network. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. https://doi.org/10.1109/CVPR.2018.00157
Chéron G, Alayrac JB, Laptev I, Schmid C (2018) A flexible model for training action localization with varying levels of supervision. Adv Neural Inf Process Syst 2018:942–953
Xia H, Zhan Y (2020) A survey on temporal action localization. IEEE Access 8:70477–70487. https://doi.org/10.1109/ACCESS.2020.2986861
Zhou ZH (2018) A brief introduction to weakly supervised learning. Natl Sci Rev 5(1):44–53. https://doi.org/10.1093/nsr/nwx106
Kolesnikov A and Lampert CH (2016) Seed, expand and constrain: three principles for weakly-supervised image segmentation. In: Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics), 9908 LNCS, pp. 695–711. https://doi.org/10.1007/978-3-319-46493-0_42
Carbonneau MA, Cheplygina V, Granger E, Gagnon G (2018) Multiple instance learning: a survey of problem characteristics and applications. Pattern Recognit 77:329–353. https://doi.org/10.1016/j.patcog.2017.10.009
Vanwinckelen G, Tragante do VO, Fierens D, Blockeel H (2016) Instance-level accuracy versus bag-level accuracy in multi-instance learning. Data Min Knowl Discov 30(2):313–341. https://doi.org/10.1007/s10618-015-0416-z
Wang L, Xiong Y, Lin D, and Van Gool L (2017) UntrimmedNets for weakly supervised action recognition and detection. In: Proc - 30th IEEE Conf Comput Vis Pattern Recognition, CVPR 2017, 2017: 6402–6411. https://doi.org/10.1109/CVPR.2017.678.
Xu Y et al (2019) Segregated temporal assembly recurrent networks for weakly supervised multiple action detection. Proc AAAI Conf Artif Intell 33:9070–9078. https://doi.org/10.1609/aaai.v33i01.33019070
Lee P, Uh Y, and Byun H (2019) Background suppression network for weakly-supervised temporal action localization. https://doi.org/10.1609/aaai.v34i07.6793
Paul S, Roy S, and Roy-Chowdhury AK (2018) W-TALC: weakly-supervised temporal activity localization and classification. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics), 11208 LNCS, pp. 588–607. https://doi.org/10.1007/978-3-030-01225-0_35
Lee P, Wang J, Lu Y, and Byun H (2020) Background modeling via uncertainty estimation for weakly-supervised action localization. pp. 1–12, [Online]. Available: http://arxiv.org/abs/2006.07006
Rashid M, Kjellstrom H, and Lee YJ (2020) Action graphs: weakly-supervised action localization with graph convolution networks. In: Proceedings - 2020 IEEE winter conference on applications of computer vision, WACV 2020, pp. 604–613. https://doi.org/10.1109/WACV45572.2020.9093404
Shi B, Dai Q, Mu Y, and Wang J (2020) Weakly-supervised action localization by generative attention modelling. pp. 1006–1016. https://doi.org/10.1109/cvpr42600.2020.00109
Schindler K and Van Gool L (2008) Action snippets: How many frames does human action recognition require?. In: 26th IEEE Conf Comput Vis Pattern Recognition, CVPR. https://doi.org/10.1109/CVPR.2008.4587730
Liu D, Jiang T, and Wang Y (2019) Completeness modeling and context separation for weakly supervised temporal action localization. In: Proceedings of the IEEE computer society conference on computer vision and pattern recognition, 2019:1298–1307. https://doi.org/10.1109/CVPR.2019.00139
Singh KK and Lee YJ (2017) Hide-and-seek: forcing a network to be meticulous for weakly-supervised object and action localization. In: Proc IEEE Int Conf Comput Vis, 2017: 3544–3553. https://doi.org/10.1109/ICCV.2017.381
Shou Z, Gao H, Zhang L, Miyazawa K, and Chang SF (2018) AutoLoc: weakly-supervised temporal action localization in untrimmed videos. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics), vol. 11220 LNCS, pp. 162–179. https://doi.org/10.1007/978-3-030-01270-0_10
Liu Z et al (2019) Weakly supervised temporal action localization through contrast based evaluation networks. Proc IEEE Int Conf Comput Vis. https://doi.org/10.1109/ICCV.2019.00400
Zeng R, Gan C, Chen P, Huang W, Wu Q, Tan M (2019) Breaking winner-takes-all: iterative-winners-out networks for weakly supervised temporal action localization. IEEE Trans Image Process 28(12):5797–5808. https://doi.org/10.1109/TIP.2019.2922108
Su H, Zhao X, and Lin T (2019) Cascaded pyramid mining network for weakly supervised temporal action localization. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics), vol. 11362 LNCS, pp. 558–574. https://doi.org/10.1007/978-3-030-20890-5_36
Su H, Zhao X, Lin T, and Fei H (2018) Weakly supervised temporal action detection with shot-based temporal pooling network. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics), 11304 LNCS, pp. 426–436. https://doi.org/10.1007/978-3-030-04212-7_37
Russakovsky O et al (2015) ImageNet large scale visual recognition challenge. Int J Comput Vis 115(3):211–252. https://doi.org/10.1007/s11263-015-0816-y
Kay W et al (2017) The kinetics human action video dataset, [Online]. Available: http://arxiv.org/abs/1705.06950
Zach C, Pock T, and Bischof H (2007) A duality based approach for realtime TV-L1 optical flow. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics), vol. 4713 LNCS, pp. 214–223. https://doi.org/10.1007/978-3-540-74936-3_22
Soomro K, Zamir AR, and Shah M (2012) UCF101: a dataset of 101 human actions classes from videos in the wild, [Online]. Available: http://arxiv.org/abs/1212.0402
Kuehne H, Jhuang H, Garrote E, Poggio T, Serre T (2011) HMDB: a large video database for human motion recognition. Proc IEEE Int Conf Comput Vis. https://doi.org/10.1109/ICCV.2011.6126543
Simonyan K, Zisserman A (2014) Two-stream convolutional networks for action recognition in videos. Adv Neural Inf Process Syst 1(January):568–576
Wang L et al (2016) Temporal segment networks: Towards good practices for deep action recognition. Lect Notes Comput Sci (including Subser. Lect Notes Artif Intell Lect Notes Bioinformatics), vol. 9912 LNCS, pp. 20–36. https://doi.org/10.1007/978-3-319-46484-8_2
Karpathy A, Toderici G, Shetty S, Leung T, Sukthankar R, Li FF (2014) Large-scale video classification with convolutional neural networks. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. https://doi.org/10.1109/CVPR.2014.223
Feichtenhofer C, Pinz A, Zisserman A (2016) Convolutional two-stream network fusion for video action recognition. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. https://doi.org/10.1109/CVPR.2016.213
Dai X, Singh B, Zhang G, Davis LS, Chen YQ (2017) Temporal context network for activity localization in videos. Proc IEEE Int Conf Comput Vis. https://doi.org/10.1109/ICCV.2017.610
Zhong JX, Li N, Kong W, Zhang T, Li TH, and Li G (2018) Step-by-step erasion, one-by-one collection: a weakly supervised temporal action detector. In: MM 2018 - Proceedings of the 2018 ACM multimedia conference, no. 2014, pp. 35–44. https://doi.org/10.1145/3240508.3240511
Huang L, Huang Y, Ouyang W, and Wang L (2020) Relational prototypical network for weakly supervised temporal action localization. Aaai
Carreira J and Zisserman A (2017) Quo Vadis, action recognition? A new model and the kinetics dataset. In: Proc. - 30th IEEE Conf Comput Vis Pattern Recognition, CVPR 2017, 2017: 4724–4733. https://doi.org/10.1109/CVPR.2017.502
Ioffe S and Szegedy C (2015) Batch normalization: accelerating deep network training by reducing internal covariate shift. In: 32nd Int Conf Mach Learn. ICML 2015, 1:448–456
Nguyen P, Ramanan D, Fowlkes C (2019) Weakly-supervised action localization with background modeling. Proc IEEE Int Conf Comput Vis. https://doi.org/10.1109/ICCV.2019.00560
Kang Z, Wang L, Liu Z, Zhang Q, Zheng N (2019) Extracting action sensitive features to facilitate weakly-supervised action localization. IFIP Adv Inform Commun Technol. https://doi.org/10.1007/978-3-030-19823-7_15
Zhai Y, Wang L, Liu Z, Zhang Q, Hua G, Zheng N (2019) Action coherence network for weakly supervised temporal action localization. Proc - Int Conf Image Process. https://doi.org/10.1109/ICIP.2019.8803447
Zhang C et al (2019) Adversarial seeded sequence growing for weakly-supervised temporal action localization. In: MM 2019 - Proc 27th ACM Int Conf Multimed, pp. 738–746. https://doi.org/10.1145/3343031.3351044
Yuan Y, Lyu Y, Shen X, Tsang IW, and Yeung DY (2019) Marginalized average attentional network for weakly-supervised learning. In: 7th Int Conf Learn. Represent. ICLR 2019, pp. 1–19
Min K and Corso JJ (2020) Adversarial background-aware loss for weakly-supervised temporal activity localization. ECCV 2020, [Online]. Available: http://arxiv.org/abs/2007.06643
Nair V and Hinton GE (2010) Rectified linear units improve restricted Boltzmann machines. In: ICML 2010 - Proceedings, 27th Int Conf Mach Learn, pp. 807–814
Zhou B, Khosla A, Lapedriza A, Oliva A, Torralba A (2016) Learning deep features for discriminative localization. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. https://doi.org/10.1109/CVPR.2016.319
Narayan S, Cholakkal H, Hayat M, Khan FS, Yang MH, and Shao L (2020) D2-Net: weakly-supervised action localization via discriminative embeddings and denoised activations. arXiv, no. December
Islam A and Radke RJ (2020) Weakly supervised temporal action localization using deep metric learning. In: Proceedings - 2020 IEEE winter conference on applications of computer vision, WACV 2020, pp. 536–545. https://doi.org/10.1109/WACV45572.2020.9093620
Idrees H et al (2017) The THUMOS challenge on action recognition for videos ‘in the wild.’ Comput Vis Image Underst 155:1–23. https://doi.org/10.1016/j.cviu.2016.10.018
Heilbron FC, Escorcia V, Ghanem B, Niebles JC (2015) ActivityNet: a large-scale video benchmark for human activity understanding. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. https://doi.org/10.1109/CVPR.2015.7298698
Sigurdsson GA, Varol G, Wang X, Farhadi A, Laptev I, and Gupta A (2016) Hollywood in homes: crowdsourcing data collection for activity understanding. Lect Notes Comput Sci (including Subser. Lect Notes Artif Intell Lect Notes Bioinformatics), vol. 9905 LNCS, pp. 510–526. https://doi.org/10.1007/978-3-319-46448-0_31
Zhao H, Torralba A, Torresani L, Yan Z (2019) HACS: Human action clips and segments dataset for recognition and temporal localization. Proc IEEE Int Conf Comput Vis. https://doi.org/10.1109/ICCV.2019.00876
Huang Z, Wang X, Wang JJ, Liu W, and Wang JJ (2018) Weakly-supervised semantic segmentation network with deep seeded region growing. In: Proceedings of the IEEE computer society conference on computer vision and pattern recognition, pp. 7014–7023. https://doi.org/10.1109/CVPR.2018.00733
Islam A, Long C, and Radke RJ (2021) A hybrid attention mechanism for weakly-supervised temporal action localization, no. Mil, [Online]. Available: http://arxiv.org/abs/2101.00545
Ge Y, Qin X, Yang D, Jagersand M (2021) Deep snippet selective network for weakly supervised temporal action localization. Pattern Recognit 110:107686. https://doi.org/10.1016/j.patcog.2020.107686
Yu T, Ren Z, Li Y, Yan E, Xu N, Yuan J (2019) Temporal structure mining for weakly supervised action detection. Proc IEEE Int Conf Comput Vis. https://doi.org/10.1109/ICCV.2019.00562
Hendrycks D and Gimpel K (2016) A baseline for detecting misclassified and out-of-distribution examples in neural networks. 5th Int Conf Learn Represent ICLR 2017 – Conf Track Proc, pp. 1–12. [Online]. Available: http://arxiv.org/abs/1610.02136
Hou R, Sukthankar R, and Shah M (2017) Real-time temporal action localization in untrimmed videos by sub-action discovery. Br Mach Vis Conf, BMVC . https://doi.org/10.5244/c.31.91
Heidarivincheh F, Mirmehdi M, and Damen D (2019) Weakly-supervised completion moment detection using temporal attention. Proc. - 2019 Int Conf Comput Vis Work. ICCVW 2019, pp. 1188–1196. https://doi.org/10.1109/ICCVW.2019.00150
Luo Z et al (2020) Weakly-supervised action localization with expectation-maximization multi-instance learning. Lect. Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics), 12374 LNCS, no. Mil, pp. 729–745. https://doi.org/10.1007/978-3-030-58526-6_43
Zhang XY, Li C, Shi H, Zhu X, Li P, Dong J (2020) AdapNet: adaptability decomposing encoder-decoder network for weakly supervised action recognition and localization. IEEE Trans Neural Netw Learn Syst. https://doi.org/10.1109/TNNLS.2019.2962815
Snell J, Swersky K, Zemel R (2017) Prototypical networks for few-shot learning. Adv Neural Inform Process Syst 2017:4078–4088
Kingma DP and J. L. Ba (2015) Adam: a method for stochastic optimization, 3rd Int Conf Learn Represent. ICLR 2015 - Conf Track Proc, pp. 1–15
Zhao Y, Xiong Y, Wang L, Wu Z, Tang X, Lin D (2020) Temporal action detection with structured segment networks. Int J Comput Vis 128(1):74–95. https://doi.org/10.1007/s11263-019-01211-2
Defferrard M, Bresson X, and Vandergheynst P (2016) Convolutional neural networks on graphs with fast localized spectral filtering. Adv Neural Inform Process Syst, no. Nips, pp. 3844–3852
Pang J, Cheung G (2017) Graph laplacian regularization for image denoising: analysis in the continuous domain. IEEE Trans Image Process 26(4):1770–1785. https://doi.org/10.1109/TIP.2017.2651400
Zhai Y, Wang L, Tang W, Zhang Q, and Yuan J (2020) Two-stream consensus network for weakly-supervised temporal action localization. In: Proc Eur. Conf Comput Vis, no. Mil, pp. 1–17
Gong G, Wang X, Mu Y, Tian Q (2020) Learning temporal co-attention models for unsupervised video action localization. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. https://doi.org/10.1109/CVPR42600.2020.00984
Acknowledgements
This work has been supported in part by the Ministry of Higher Education Malaysia for Fundamental Research Grant Scheme with Project Code: FRGS/1/2019/ICT02/USM/02/1.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Baraka, A., Mohd Noor, M.H. Weakly-supervised temporal action localization: a survey. Neural Comput & Applic 34, 8479–8499 (2022). https://doi.org/10.1007/s00521-022-07102-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-022-07102-x