Combining embedded accelerometers with computer vision for recognizing food preparation activities
The DOI matches its Crossref record; 32 of the 33 references it cites were found in a registry or an index.
Abstract
This paper introduces a publicly available dataset of complex activities that involve manipulative gestures. The dataset captures people preparing mixed salads and contains more than 4.5 hours of accelerometer and RGB-D video data, detailed annotations, and an evaluation protocol for comparison of activity recognition algorithms. Providing baseline results for one possible activity recognition task, this paper further investigates modality fusion methods at different stages of the recognition pipeline: (i) prior to feature extraction through accelerometer localization, (ii) at feature level via feature concatenation, and (iii) at classification level by combining classifier outputs. Empirical evaluation shows that fusing information captured by these sensor types can considerably improve recognition performance.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
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
“The 50Salads dataset contains 50 videos with 17 action classes.”
From MS-TCN++: Multi-Stage Temporal Convolutional Network for Action Segmentation (2023) · cites this paper as [25]
In the citing paper
“Our model achieves state-of-the-art results on three challenging datasets: 50Salads, Georgia Tech Egocentric Activities (GTEA), and the Breakfast dataset.”
From MS-TCN++: Multi-Stage Temporal Convolutional Network for Action Segmentation (2023) · cites this paper as [25]
Citations
Cites
A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection
Kohavi, R. (1995) · Zenodo · doi:10.5281/zenodo.19712698 · Cited by 10,909
Kittler, J., Hatef, M., Duin, R. P. W., & Matas, J. (1998) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/34.667881 · Cited by 5,290 · cited 2×
“For classifier fusion we tried the sum-rule and the product-rule [11] as well as Random Decision Forests as a non-linear combination method.”
MethodRecognizing human actions: A local SVM approach
Schüldt, C., Laptev, I., & Barbara Caputo, B. (2004) · Cited by 2,906
Laptev, I. (2005) · International Journal of Computer Vision · doi:10.1007/s11263-005-1838-7 · Cited by 2,629
Action recognition by dense trajectories
Wang, H., Klaser, A., Schmid, C., & Liu, C.-L. (2011) · CVPR 2011 · doi:10.1109/cvpr.2011.5995407 · Cited by 2,154 · cited 3×
“In addition to yielding low recognition performance, extracting local descriptors around densely sampled locations as proposed in [26] is computationally demanding in real-time, e.”
BackgroundMethodHuman activity analysis: A review
Aggarwal, J. K., & Ryoo, M. S. (2011) · ACM Computing Surveys · doi:10.1145/1922649.1922653 · Cited by 2,056
Marszalek, M., Laptev, I., & Schmid, C. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206557 · Cited by 1,168 · cited 3×
“, KTH [23], YouTube [14], Hollywood2 [15] and URADL [17]).”
BackgroundRecognizing realistic actions from videos “in the wild”
Liu, J., Luo, J., & Shah, M. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206744 · Cited by 1,075
Collecting complex activity datasets in highly rich networked sensor environments
Roggen, D., Calatroni, A., Rossi, M., Holleczek, T., Forster, K., Troster, G., Lukowicz, P., Bannach, D., Pirkl, G., Ferscha, A., et al. (2010) · 2010 Seventh International Conference on Networked Sensing Systems (INSS) · doi:10.1109/inss.2010.5573462 · Cited by 788
Criminisi, A. (2011) · now publishers, Inc. eBooks · doi:10.1561/9781601985415 · Cited by 703
Research works printed but not linked: 6
- 10.5555/2029024 · doi:10.5555/2029024
- 10.5555/2283516.2283683 · doi:10.5555/2283516.2283683
- M. E. Pollack . Intelligent technology for an aging population: The use of AI to assist elders with cognitive impairment . AI Magazine , 26 ( 2 ), 2005 . M. E. Pollack. Intelligent technology for an aging population: The use of AI to assist elders with cognitive impairment. AI Magazine, 26(2), 2005.
- 10.5555/2354409.2354909 · doi:10.5555/2354409.2354909
- 10.5555/1018429.1020906 · doi:10.5555/1018429.1020906
- 10.5555/1786014.1786017 · doi:10.5555/1786014.1786017
Cited by
Items 201–300 of 447
Predicting Human Intentions from Motion Cues Only: A 2D+3D Fusion Approach
Zunino, A., Cavazza, J., Koul, A., Cavallo, A., Becchio, C., & Murino, V. (2017) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3123266.3123298 · Cited by 15
“For instance, MPII-CAD [44] and Salad 50 [53] cover more articulated (cooking) actions, while UCF-101 [50] and HDMB51 [32] collect YouTube videos, thus guaranteeing a broad variability of backgrounds and context.”
BackgroundWeber, J., Esau, M., Schiller, M. R. G., Thaden, E., Manstetten, D., & Stevens, G. (2023) · Designing Interactive Systems Conference · doi:10.1145/3563657.3595975 · Cited by 15
“…assessment confrms the viability of the chosen combination of modalities for “Cookie”, but does not preclude potential improvements obtained by the inclusion of gestures or connected/instrumented cooking utensils (cf. [ 53, 63]) into the interaction concept as part of possible future work.”
BackgroundHousehold Surface Interactions: Understanding User Input Preferences and Perceived Home Experiences
Tigwell, G. W., & Crabb, M. (2020) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3313831.3376856 · Cited by 12
“Computer vision [45] and tag systems [57] have previously been used to bridge the gap between digital and physical worlds; this should be understood by HSI systems to reduce unintended input.”
BackgroundFrom Human to Robot Everyday Activity
Mason, C., Gadzicki, K., Meier, M., Ahrens, F., Kluß, T., C., J. L. M., Putze, F., Fehr, T., Zetzsche, C., Herrmann, M., et al. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9340706 · Cited by 14
Das, P., & Ortega, A. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9413833 · Cited by 12
Zhong, C., Reibman, A. R., Mina, H. A., & Deering, A. J. (2021) · Journal of Imaging · doi:10.3390/jimaging7090170 · Cited by 16
“The MPII cooking activity dataset [57], 50salads dataset [58], and Breakfast Actions Dataset [59] recorded food handling and cooking in a home scenario.”
Weakly-Supervised Temporal Action Detection for Fine-Grained Videos with Hierarchical Atomic Actions
Li, Z., He, L., & Xu, H. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20080-9_33 · Cited by 18
Liu, S., Mangla, T., Shaowang, T., Zhao, J., Paparrizos, J., Krishnan, S., & Feamster, N. (2023) · Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies · doi:10.1145/3580818 · Cited by 37
“Similar datasets include TUM Kitchen [57] and 50 Salads [54].”
MethodMCFM: Mutual Cross Fusion Module for Intermediate Fusion-Based Action Segmentation
Ishihara, K., Nakano, G., & Inoshita, T. (2022) · Proceedings - International Conference on Image Processing · doi:10.1109/icip46576.2022.9897444 · Cited by 12
Recognising complex activities with histograms of relative tracklets
Stein, S., & McKenna, S. J. (2017) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2016.08.012 · Cited by 11
Self-citationVideo Action Segmentation via Contextually Refined Temporal Keypoints
Jiang, B., Jin, Y., Tan, Z., & Mu, Y. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01272 · Cited by 18
“The metric (b) is reasonable under 50salads [38] because there are few actions with a total duration less than 10 frames.”
BackgroundMethodInfluentialCao, C., Sun, Z., Lv, Q., Min, L., & Zhang, Y. (2024) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2024.3425598 · Cited by 12
“Long-term anticipation methods are usually tested on scripted third-person datasets [18], [19], [21], [22], [23], [24], [25], [26], such as 50 Salads [27] and Breakfast [28].”
Gated Temporal Diffusion for Stochastic Long-Term Dense Anticipation
Zatsarynna, O., Bahrami, E., Farha, Y. A., Francesca, G., & Gall, J. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73001-6_26 · Cited by 20
Shuvo, M. R., Mekala, M. S., & Elyan, E. (2025) · Cognitive Computation · doi:10.1007/s12559-025-10513-2 · Cited by 17
“Both achieved superior performance with fewer parameters on datasets like 50salads [143], GTEA [144], and Breakfast [145].”
Skeleton-based Tai Chi action segmentation using trajectory primitives and content
Xu, L., Wang, Q., Lin, X.-T., Yuan, L., & Ma, X. (2022) · Neural computing & applications (Print) · doi:10.1007/s00521-022-08185-2 · Cited by 12
“Most related studies focus on video-based action segmentation [11–14], and there are also some untrimmed videos datasets [15–17] for this task.”
BackgroundOTAS: Unsupervised Boundary Detection for Object-Centric Temporal Action Segmentation
Li, Y., Xue, Z., & Xu, H. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00631 · Cited by 23
“50Salads [39] includes altogether 4 hours videos of mixing salad, whose average length is 5 minutes.”
BackgroundRecognition of ongoing complex activities by sequence prediction over a hierarchical label space
Li, W., & Fritz, M. (2016) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2016.7477586 · Cited by 13
“In contrast to this, there are also long-duration datasets collected from surveillance video like VIRAT [26] or daily living recordings like TUM Kitchen [38], CMU-MMAC [8], URDAL [24], TRECVID [27], ego-centric datasets [11, 36], and MPII Cooking [30].”
BackgroundUncertainty-Boosted Robust Video Activity Anticipation
Qi, Z., Wang, S., Zhang, W., & Huang, Q. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3393730 · Cited by 41
“As shown in Tables VII and X, our framework significantly improves the anticipation performance on RULSTM and DCR backbones in the model ensemble setting.”
MethodResultTowards an autonomous clinical decision support system
Gershov, S., Raz, A., Karpas, E., & Laufer, S. (2023) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2023.107215 · Cited by 12
Toward Computer Vision Systems That Understand Real-World Assembly Processes
Jones, J. D., Hager, G. D., & Khudanpur, S. P. (2019) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2019.00051 · Cited by 12
“For instance, people try to make a salad in[20] or build a toy airplane in [22].”
BackgroundLan, B., Li, P., Yin, J., Song, Y., Wang, G., Ding, H., Han, J., & Wang, F. (2025) · Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies · doi:10.1145/3749521 · Cited by 27
“However, these datasets are not well-suited for summarizing indoor daily activities due to several limitations: (1) Task-Specific Datasets: many datasets, such as YouCook2 [92], Breakfast Dataset [25], and 50Salads [48], focus on specific tasks like cooking or food preparation.”
Machine learning for video event recognition
Avola, D., Cascio, M., Cinque, L., Foresti, G. L., & Pannone, D. (2021) · Integrated Computer-Aided Engineering · doi:10.3233/ica-210652 · Cited by 9
Reducing the Label Bias for Timestamp Supervised Temporal Action Segmentation
Liu, K., Li, Y., Liu, S., Tan, C., & Shao, Z. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00629 · Cited by 15
“Following the previous works [17, 32, 51], we evaluate our approach on three datasets: 50Salads [36], Breakfast [19], and Georgia Tech Egocentric Activities (GTEA) [8].”
MethodLiu, Y., Zhong, X., Zhai, S., Du, Z., Gao, Z., Huang, Q., Zhang, C. Y., Jiang, B., Pandey, V. K., Han, S., et al. (2023) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2023.107672 · Cited by 10
Coherent Temporal Synthesis for Incremental Action Segmentation
Ding, G., Golong, H., & Yao, A. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.02691 · Cited by 14
“Note other TAS datasets such as GTEA [13] and 50Salads [48] are unsuitable because most of their videos share a common set of actions, making it challenging to implement a meaningful number of incremental tasks without actions overlapping.”
Cabrera-Quirós, L., & Hung, H. (2016) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/2964284.2967224 · Cited by 9
Weakly supervised action segmentation with effective use of attention and self-attention
Ng, Y. B., & Fernando, B. (2021) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2021.103298 · Cited by 11
“In this section we evaluate the performance of our model (Figure 2) on weakly supervised action segmentation task us-ing Breakfast (Kuehne et al., 2014) and 50Salad datasets(Stein and McKenna, 2013).”
MethodSayed, S., Ghoddoosian, R., Trivedi, B., & Athitsos, V. (2023) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw59228.2023.00315 · Cited by 16
“In our experiments, in addition to (3+1)ReC, we have used three public datasets commonly used for evaluating action segmentation methods: 1) The 50Salads dataset [41] contains 50 videos and 17 fine-grained action classes.”
BackgroundMethodInfluentialVLMAH: Visual-Linguistic Modeling of Action History for Effective Action Anticipation
Manousaki, V., Bacharidis, K., Papoutsakis, K. E., & Argyros, A. (2023) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw60793.2023.00206 · Cited by 16
“For 50Salads [53] we utilized pre-extracted I3D features from [10, 42], which correspond to frame-wise [1 × 2048] feature vectors, which were restructured in the form described for the ones of the Assembly dataset.”
InfluentialWang, T., Chen, K., Zheng, Z., Guo, J., Zhao, X., & Zhang, S. (2024) · Sensors · doi:10.3390/s24103241 · Cited by 10
“In the experiments, we utilized a dataset consisting of image and video files, totaling 656 files, with the average length of each video ranging between 6 and 7 min [33,34].”
MethodRecognizing Manipulation Actions from State-TransformationsPreprint
Bakr, N. A., Crowley, J. L., & Ronfard, R. P. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.05147 · Cited by 12
“For example, cut tomato and cut cucumber can be considered as two different classes as in [16].”
BackgroundTwo-Stage Active Learning for Efficient Temporal Action Segmentation
Su, Y., & Elhamifar, E. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72970-6_10 · Cited by 15
“1) 50Salads [107] consists of 50 videos, with 17 actions that depict individuals making salads.”
Friends Across Time: Multi-Scale Action Segmentation Transformer for Surgical Phase Recognition
Zhang, B., Meng, J., Cheng, B., Biskup, D., Petculescu, S., & Chapman, A. (2024) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc53108.2024.10782887 · Cited by 8
“(4 ) To demonstrate the robustness and wide applicability of our method, we evaluate our MS-AST on the 50Salads [16] and GTEA [17] datasets, which are non-medical datasets that are widely used in the video action segmentation domain.”
Action Anticipation from Soccernet Football Video Broadcasts
Dalal, M., Xarles, A., Cioppa, A., Giancola, S., Droogenbroeck, M. V., Ghanem, B., Clapés, A., Escalera, S., & Moeslund, T. B. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00605 · Cited by 16
“Action anticipation datasets are typically adapted from action localization or segmentation datasets, often focusing on daily activities, such as cooking [8, 30, 31, 44], or sourced from movie and TV-series datasets [9, 25].”
Recent Advances in Video Action Recognition with 3D Convolutions
Hara, K. (2020) · IEICE Transactions on Fundamentals of Electronics Communications and Computer Sciences · doi:10.1587/transfun.2020imp0012 · Cited by 11
“Although many video datasets exist for other video recognition tasks, such as spatio–temporal action detection [73], action segmentation [74], video caption-HARA ing [75], learning text-video embeddings [76], and understanding human–object relationships [77], we only discuss datasets used for…”
BackgroundSG-TCN: Semantic Guidance Temporal Convolutional Network for Action Segmentation
Zhang, Y., Ren, K., Zhang, C., & Yan, T. (2022) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn55064.2022.9891932 · Cited by 8
“To validate our research, we evaluate the performance of the SG-TCN on three challenging datasets: GTEA [8], 50Salads [24], Breakfast [13] and compare our proposed approach with other popular approaches.”
MethodTimestamp-Supervised Action Segmentation from the Perspective of Clustering
Du, D., Li, E., Si, L., Xu, F., & Sun, F. (2023) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2023/77 · Cited by 13
SkatingVerse: A large‐scale benchmark for comprehensive evaluation on human action understanding
Gan, Z., Jin, L., Cheng, Y., Cheng, Y., Teng, Y., Li, Z., Li, Y., Yang, W., Zhu, Z. D., Xing, J., et al. (2024) · IET Computer Vision · doi:10.1049/cvi2.12287 · Cited by 9
Multi-modal temporal action segmentation for manufacturing scenarios
Romeo, L., Marani, R., Perri, A. G., & Gall, J. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.110320 · Cited by 12
Action recognition: A comprehensive survey of tasks, methods, and challenges
Heo, S., Moon, J., & Jung, S. K. (2025) · ICT Express · doi:10.1016/j.icte.2025.11.015 · Cited by 12
Self-Attention Based Action Segmentation Using Intra-And Inter-Segment Representations
Patsch, C., & Steinbach, E. G. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10096960 · Cited by 7
“The 50Salads dataset [19] has 19 action classes and contains salad preparation videos executed by 25 subjects in a kitchen environment.”
BackgroundHOI-aware Adaptive Network for Weakly-supervised Action Segmentation
Zhang, R., Wang, S.-C., Duan, Y., Tang, Y.-S., Zhang, Y., & Tan, Y.-P. (2023) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2023/191 · Cited by 11
“We conduct our experiments on two real-world instructional video datasets: Breakfast [Kuehne et al. , 2014] and 50Salads [Stein and McKenna, 2013].”
Markov Game Video Augmentation for Action Segmentation
Aziere, N., & Todorović, S. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01242 · Cited by 9
“Datasets include three benchmarks: Breakfast [22], GTEA [12], and 50Salads [36].”
MethodOrdered Atomic Activity for Fine-grained Interactive Traffic Scenario Understanding
Agarwal, N., & Chen, Y. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00792 · Cited by 9
“Existing works have concentrated on cooking activities [37, 4, 17, 68] where object movements are limited or only a single action occurs at a given point in time.”
BackgroundLong-Term Action Anticipation Based on Contextual Alignment
Patsch, C., Zhang, J., Wu, Y., Zakour, M., Salihu, D., & Steinbach, E. G. (2024) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp48485.2024.10445978 · Cited by 7
“• We evaluate our approach based on the Breakfast [14] and the 50Salads [15] datasets, where we achieve state-of-the-art results.”
MethodRandom Walks for Temporal Action Segmentation with Timestamp Supervision
Hirsch, R., Cohen, R., Golany, T., Freedman, D., & Rivlin, E. (2024) · IEEE Workshop/Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00648 · Cited by 4
Global Spatial-Temporal Information Encoder-Decoder Based Action Segmentation in Untrimmed Video
Liu, Y., Sun, Y., Chen, Z., Feng, C., & Zhu, K. (2024) · Tsinghua Science & Technology · doi:10.26599/tst.2024.9010041 · Cited by 7
Turning to a Teacher for Timestamp Supervised Temporal Action Segmentation
Zhao, Y., & Song, Y. (2022) · IEEE International Conference on Multimedia and Expo · doi:10.1109/icme52920.2022.9859626 · Cited by 7
“We evaluate our method on three public datasets, namely Georgia Tech Egocentric Activities (GTEA) [19], 50Salads [20] and Breakfast [14].”
Manual assembly actions segmentation system using temporal-spatial-contact features
Kang, Z., Cui, J., & Chu, Z.-Y. (2023) · Robotic Intelligence and Automation · doi:10.1108/ria-01-2023-0008 · Cited by 9
Human Action Recognition Using Fusion of Depth and Inertial Sensors
Fuad, Z., & Ünel, M. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-93000-8_42 · Cited by 8
Alati, E., Mauro, L., Ntouskos, V., & Pirri, F. (2019) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2019.8803155 · Cited by 8
“Like them, we use Breakfast [14] for forecasting, though we do not use 50Salads [26], using instead both MPII [24] and CAD120 [13].”
MethodLearning to Abstract and Predict Human Actions
de Morais, R. F. A. B., Le, V., Tran, T., & Venkatesh, S. (2020) · British Machine Vision Conference (BMVC) · doi:10.5244/c.34.195 · Cited by 7
Datasets on object manipulation and interaction: a surveyPreprint
Huang, Y., & Sun, Y. M. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1607.00442 · Cited by 7
“4) timestamp: [7], [14] 5) implicit: [13], [15], [18].”
BackgroundMethodInfluentialEnd-to-End Streaming Video Temporal Action Segmentation With Reinforcement Learning
Zhang, J., Wen, W., Liu, S., Huang, G. M., Li, Y., Li, Q., & Feng, L. (2025) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2025.3550910 · Cited by 7
“The 50Salads [39] dataset contains 50 videos Algorithm 3 Monte Carlo Episodic REINFORCE Learning for STAS Task (MC) with 17 action classes.”
The BRIO-TA Dataset: Understanding Anomalous Assembly Process in Manufacturing
Moriwaki, K., Nakano, G., & Inoshita, T. (2022) · Proceedings - International Conference on Image Processing · doi:10.1109/icip46576.2022.9897369 · Cited by 6
“BRIO-TA differs from existing datasets [4, 5, 6] in that it consists of various work processes for the same type of work.”
BackgroundXu, X., Li, Y., & Lu, C. (2023) · International Journal of Computer Vision · doi:10.1007/s11263-023-01850-6 · Cited by 9
“Observation In Video Action Anticipation, we set observation time τ o = 10s for EPIC-KITCHENS (Damen et al., 2021, 2018) series and 50S (Stein & McKenna , 2013), but τ o = 5s for EG+ (Li et al., 2018).”
BackgroundMethodInfluentialU-Transformer-based multi-levels refinement for weakly supervised action segmentation
Ke, X., Miao, X., & Guo, W.-Z. (2023) · Pattern Recognition · doi:10.1016/j.patcog.2023.110199 · Cited by 7
Semantically Guided Representation Learning For Action Anticipation
Diko, A., Avola, D., Prenkaj, B., Fontana, F., & Cinque, L. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73390-1_26 · Cited by 18
“Moreover, we evaluate S-GEAR on an exocentric dataset, namely 50 Salads [27], to demonstrate its versatility in tackling long-term dense anticipation.”
BackgroundMethodLLMAction: Adapting Large Language Model for Long-Term Action Anticipation
Wang, B., Tian, Y., Wang, C., & Yang, L. (2024) · Lecture notes in computer science · doi:10.1007/978-981-97-8792-0_19 · Cited by 9
Learning alternative ways of performing a task
Nieves, D., Ramírez-Quintana, M. J., Monserrat, C., Ferri, C., & Hernández‐Orallo, J. (2020) · Expert Systems with Applications · doi:10.1016/j.eswa.2020.113263 · Cited by 6
“Other datasets in the cooking domain, such as the salad preparation dataset (Stein & McKenna, 2013) or the sandwich preparation dataset (Spriggs et al., 2009), present a huge variability in executions, not really displaying a short number of patterns, and applying pattern extraction to them would…”
BackgroundEfficient Two-Stream Network for Online Video Action Segmentation
Kang, M., Park, R., & Park, H. (2022) · IEEE Access · doi:10.1109/access.2022.3201208 · Cited by 7
“We evaluate the performance of our proposed pipelines on public action segmentation datasets: 50Salads [39] and GTEA [38].”
MethodMyers, M. K., Wright, N., McGough, A. S., & Martin, N. (2023) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw58289.2023.00052 · Cited by 6
“Compared to other action segmentation datasets ours is much more fine-grained than large datasets such as Breakfast [20] (77 hours - 10 actions) and larger in size than comparative fine-grained datasets such as GTEA [7] (28 minutes - 20 actions) and 50 salads [31] (4 hours - 17 actions).”
BackgroundElangovan, N., de Godoy, R. V., Sanches, F. P., Wang, K., White, T., Jarvis, P., & Liarokapis, M. V. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161171 · Cited by 8
“…some available datasets focus on recording subjects performing a specific type o kitchen task, such as setting the table [12] or preparing a meal [13], [14], recent works are trying to capture a more significant sampling of tasks, in order to represent the dynamic nature of the kitchen…”
BackgroundThe effects of selected object features on a pick-and-place task: A human multimodal dataset
Lastrico, L., Belcamino, V., Carfì, A., Vignolo, A., Sciutti, A., Mastrogiovanni, F., & Rea, F. (2023) · The International Journal of Robotics Research · doi:10.1177/02783649231210965 · Cited by 6
“…al. 2016), most of the time focusing on the interaction between the humans and the objects in specific applications, e.g., activities of daily living (Huang and Sun 2019), kitchen-related actions (Tenorth et al. 2009; Stein and Mckenna 2013; Nicora et al. 2020), or handovers (Carf`ı et al. 2019).”
BackgroundRecognition and localization of food in cooking videos
Bakr, N. A., Ronfard, R. P., & Crowley, J. L. (2018) · doi:10.1145/3230519.3230590 · Cited by 6
“For testing, we extracted key-frames from the 50 salads dataset [11].”
MethodMeier, M., Mason, C., Putze, F., & Schultz, T. (2019) · Interspeech (USB) · doi:10.21437/interspeech.2019-3072 · Cited by 8
“The 50 Salads dataset [7] was produced in a non-natural kitchen setting using a third person top-down point of view camera, where participants prepared a meal according to a recipe.”
MethodEnhancing Temporal Action Localization with Transfer Learning from Action Recognition
Iqbal, A., Richard, A., & Gall, J. (2019) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw.2019.00191 · Cited by 8
“One is focusing on long-range temporal modeling of actions in videos, particularly for datasets where there are clear semantical dependencies between succeeding actions such as in cooking videos [14, 27, 30].”
BackgroundDifferentiable Grammars for Videos
Piergiovanni, A., Angelova, A., & Ryoo, M. S. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i07.6861 · Cited by 6
“50 Salads: This dataset contains 50 videos of making sal-ads, e.g., cutting vegetables and mixing them.”
MethodA temporal and channel-combined attention block for action segmentation
Yang, D.-W., Cao, Z., Mao, L., & Zhang, R.-B. (2022) · Applied intelligence (Boston) · doi:10.1007/s10489-022-03569-2 · Cited by 6
“• The performance of the baseline network in the three datasets of action segmentation (GTEA [21], 50Salads [22], and Breakfast [23]) has been greatly improved by inserting our module.”
BackgroundDPcode: Privacy-Preserving Frequent Visual Patterns Publication on Cloud
Qin, Z., Ren, K., Yu, T., & Weng, J. (2016) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2016.2535729 · Cited by 6
“We evaluate the utility of DPcode over a variety of benchmark video datasets, which are described as follows: 50 Salads dataset [25] contains 25 people preparing two mixed salads each and contains over 4 h of annotated accelerometer and RGB-D video data.”
MethodRobot learning of manipulation activities with overall planning through precedence graph
Ye, X., Lin, Z., & Yang, Y. (2019) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2019.03.011 · Cited by 6
Hierarchical Attention Network for Action Segmentation
Gammulle, H., Denman, S., Sridharan, S., & Fookes, C. (2020) · Pattern Recognition Letters · doi:10.1016/j.patrec.2020.01.023 · Cited by 9
“The evaluation results for MERL Shopping (Singh et al., 2016a), 50 salads (Stein and McKenna, 2013) and Georgia Tech Egocentric (Li et al., 2015b) datasets are presented in Tables 1, 2 and 3, respectively.”
MethodInfluentialLearning Temporal Video Procedure Segmentation from an Automatically Collected Large Dataset
Ji, L., Wu, C., Zhou, D., Yan, K., Cui, E., Chen, X., & Duan, N. (2022) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv51458.2022.00279 · Cited by 12
On Evaluating Weakly Supervised Action Segmentation MethodsPreprint
Souri, Y., Richard, A., Minciullo, L., & Gall, J. (2020) · arXiv.org · doi:10.48550/arxiv.2005.09743 · Cited by 7
“It is usually studied in the context of instructional activities performed by a single person [6, 15, 17] and fully supervised approaches for action segmentation achieve already very good results [7, 10, 9, 1].”
BackgroundSelf-Supervised Learning for Temporal Action Segmentation in Industrial and Manufacturing Videos
Vybornova, Y., Aleshin, M., Illarionova, S., Novikov, I., Shadrin, D., Nikonorov, A. V., & Burnaev, E. V. (2025) · IEEE Access · doi:10.1109/access.2025.3545768 · Cited by 5
“Some datasets, such as those containing videos of routine processes such as meal preparation [1], [2], are more readily available.”
TSCL: Timestamp Supervised Contrastive Learning for Action Segmentation
Patsch, C., Wu, Y., Salihu, D., Zakour, M., & Steinbach, E. G. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3427553 · Cited by 5
“(cid:2) Our approach outperforms the state-of-the-art on Georgia Tech Egocentric Activities (GTEA) [11], 50 Salads [12] and Breakfast [13] for the F1 and Edit metrics.”
MethodFine-grained Action Detection in Untrimmed Surveillance Videos
Aakur, S. N., Sawyer, D., & Sarkar, S. (2019) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw.2019.00014 · Cited by 5
“There have been several approaches to temporal detection of actions in untrimmed videos [6, 7, 8, 9, 10] and spatial detection in trimmed videos [11, 12].”
MethodStreaming Video Temporal Action Segmentation in Real Time
Wen, W., Li, Y., Dong, Z., Feng, L., Yang, W., & Liu, S. (2023) · International Conference on Intelligent Systems and Knowledge Engineering (ISKE) · doi:10.1109/iske60036.2023.10481438 · Cited by 6
“The 50Salads [21] dataset contains 50 videos with 17 action classes.”
BackgroundMulti-Label Action Anticipation for Real-World Videos With Scene Understanding
Zhang, Y., Li, X., Xie, H., Zhuang, W., Guo, S., & Li, Z. (2024) · IEEE Transactions on Image Processing · doi:10.1109/tip.2024.3391692 · Cited by 7
“These methods are mostly targeted to the standard datasets—such as 50Salads [11], Breakfast [12], Epic-Kitchens [13], CAD-120 [14], MPII-Cooking [15] and Watch-n-patch [16]—on which they have achieved promising prediction results.”
MethodSMC-NCA: Semantic-Guided Multi-Level Contrast for Semi-Supervised Temporal Action Segmentation
Zhou, F., Jiang, Z., Zhou, H., & Li, X. (2024) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2024.3452980 · Cited by 7
“We evaluate the proposed method on three challenging datasets: 50Salads [51] (50 videos, 19 actions), GTEA [52] (28 videos, 11 actions), and Breakfast Actions [53] (1712 videos, 10 complex activities, 48 actions).”
Benchmarking and enhancing surgical phase recognition models for robot-assisted esophagectomy
Li, Y., van Jaarsveld, R. C., de Jong, R., Bongers, J., Kuiper, G. M., van Hillegersberg, R., Ruurda, J. P. H., Breeuwer, M. M., & Khalil, Y. A. (2025) · SPIE Medical Imaging: Image-Guided Procedures, Robotic Interventions, and Modeling · doi:10.1117/12.3040084 · Cited by 7
“Inspired by ASformer [17] and its success on the Breakfast [18] and 50 Salads [19] datasets, which are well-established benchmarks for temporal action segmentation, we identify parallels between these tasks and surgical phase recognition.”
Mandour, M. F., Abd-Eldaim, M. M., Abdelwahab, S., Abu-Elnaga, H. I., Elshahidy, M., Azab, A. M. H., & Eltarabily, M. M. (2013) · INFM-OAR (INFN Catania) · doi:10.5281/zenodo.19335 · Cited by 5
Yang, F., Odashima, S., Masui, S., & Jiang, S. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341931 · Cited by 5
“Breakfast [35] 50Salads [34] GTEA [30] F1@ { 10,25,50 } Edit Acc F1@ { 10,25,50 } Edit Acc F1@ { 10,25,50 } Edit Acc Baseline [25] 70 .”
BackgroundMethodInfluentialA comprehensive survey of procedural video datasets
Tan, H. L., Zhu, H., Lim, J., & Tan, C. (2020) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2020.103107 · Cited by 7
A neural network architecture for automatic and objective surgical skill assessment
Benmansour, M., Handouzi, W., & Malti, A. (2018) · International Conference on Electrical Sciences and Technologies in Maghreb · doi:10.1109/cistem.2018.8613550 · Cited by 6
Synchronized Multimodal Recording of a Table Setting Dataset
Meier, M., Mason, C., Porzel, R., Putze, F., & Schultz, T. (2018) · Cited by 5
Intention from Motion.Preprint
Zunino, A., Cavazza, J., Koul, A., Cavallo, A., Becchio, C., & Murino, V. (2016) · arXiv (Cornell University) · Cited by 5
Kusk Object Dataset: Recording access to objects in food preparation
Hashimoto, A., Mori, S., Iiyama, M., & Minoh, M. (2016) · IEEE International Conference on Multimedia & Expo Workshops (ICMEW) · doi:10.1109/icmew.2016.7574771 · Cited by 4
“50 Salads dataset [13] observes food preparation activities by cameras set on the ceiling.”
BackgroundRecognizing Plans by Learning Embeddings from Observed Action Distributions
Zha, Y., Li, Y., Gopalakrishnan, S., Li, B., & Subbarao Kambhampati, B. (2018) · Adaptive Agents and Multi-Agents Systems · doi:10.65109/pcgg9004 · Cited by 5
Cabrera-Quirós, L., & Hung, H. (2018) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2018.2888798 · Cited by 6
“tion/activity classification [11], [12] or group detection [5], among many others.”
A Literature Review: Geometric Methods and Their Applications in Human-Related Analysis
Gong, W., Zhang, B., Wang, C., Yue, H., Li, C., Xing, L., Qiao, Y., Zhang, W., & Gong, F. (2019) · Sensors · doi:10.3390/s19122809 · Cited by 4
“The RGB video data have a resolution of 640 ˆ 480 pixels at 30 Hz and the depth maps a resolution of 640 ˆ 480 pixels at 30 Hz, and the three-axis accelerometer data are at 50 Hz [153].”
MethodStacking-Based Attention Temporal Convolutional Network for Action Segmentation
Yang, L., Jiang, Y., Hong, J.-K., Wu, Z., Yang, Z., & Long, J. (2023) · IEEE International Conference on Acoustics, Speech, and Signal Processing · doi:10.1109/icassp49357.2023.10097024 · Cited by 3
“We conduct our SATC module on three challenging datasets shown in Table 1: Georgia Tech Egocentric Activities (GTEA) [17], 50Salads [18] and Breakfast [19].”
MethodUnsupervised Action Segmentation of Untrimmed Egocentric Videos
Perochon, S., & Oudre, L. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10097216 · Cited by 4
“In addition to these two datasets, we provide results for the exocentric dataset 50Salads [21], which consists in 50 videos in third-person view of 25 different actors performing a cooking activity, with a total duration of approximately 4.5 hours and 14% of background frames.”
BackgroundInfluentialFrom Recognition to Prediction: Leveraging Sequence Reasoning for Action Anticipation
Liu, X., Hao, C., Yu, Z., Yue, H., & Yang, J. (2024) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3687474 · Cited by 7
“We conducted experiments on three large-scale datasets: Epic-kitchen-100 (EK100) [14], EGTEA Gaze+ [39], and 50salads [63].”
BackgroundMethodContinuous Action Recognition and Segmentation in Untrimmed Videos
Bai, R., Zhao, Q. Q., Zhou, S., Li, Y., Zhao, X., & Wang, J. (2018) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr.2018.8546019 · Cited by 4
“1) 50Salads [19]: The dataset contains 50 untrimmed videos from 25 people, each contains the preparation process of 2 mixed salads.”
BackgroundMethodInfluentialHand Graph Representations for Unsupervised Segmentation of Complex Activities
Das, P., Kao, J.-Y., Ortega, A., Sawada, T., Mansour, H., Vetro, A., & Minezawa, A. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8683643 · Cited by 4
“Complex activities such as assembly tasks [1], food preparation [2], surgical procedures [3], etc can often be broken down into a sequence of smaller sub-tasks.”
Self-supervised temporal event segmentation inspired by cognitive theories
Mounir, R., Aakur, S. N., & Sarkar, S. (2022) · Elsevier eBooks · doi:10.1016/b978-0-12-822109-9.00021-7 · Cited by 6
Modelling daily actions through hand-based spatio-temporal features
Mur, O., Frigola, M., & Casals, A. (2015) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar.2015.7251499 · Cited by 4
“Stain and McKenna [8] combine embedded accelerometers with computer vision for recognizing food preparation.”
BackgroundMethodInfluentialLow-rank Random Tensor for Bilinear Pooling.Preprint
Zhang, Y., Muandet, K., Ma, Q., Heiko Neumann, B., & Tang, S. (2019) · arXiv (Cornell University) · Cited by 4
Unsupervised Action Segmentation for Instructional VideosPreprint
Piergiovanni, A., Angelova, A., Ryoo, M. S., & Essa, I. A. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.03738 · Cited by 4
“We evaluate our unsupervised atomic action discovery approach on multiple video segmentation datasets: (1) 50salads dataset [16], which contains 50 videos of people making salads (i.”
Metadata
- Authors
- Stein, S., & McKenna, S. J.
- Affiliations
- University of Dundee (GB)
- Venue
- Proceedings of the 2013 ACM international joint conference on Pervasive and ubiquitous computing
- Publisher
- ACM
- Published
- 2013-09-08
- Pages
- 729-738
- OpenAlex
- W2109698606
- Cited by
- 448
- References
- 33
- Citations by year
- 2026: 26 · 2025: 41 · 2024: 56 · 2023: 56 · 2022: 54 · 2021: 50 · 2020: 37 · 2019: 49 · 2018: 26 · 2017: 15 · 2016: 24 · 2015: 7 · 2014: 5 · 2013: 2
- FWCI
- 8.08
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) (publishedVersion)
- Discovery Research Portal (University of Dundee) (submittedVersion)
- ENLIGHTEN (Jurnal Bimbingan dan Konseling Islam) (submittedVersion)
- Funders
- Research Councils UK
- Engineering and Physical Sciences Research Council (EP/G066019/1)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
- Physical Sciences › Computer Science › Human-Computer Interaction › Hand Gesture Recognition Systems
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Human Pose and Action Recognition
- Keywords
- activity recognition, embedded accelerometers, RGB-D video, multimodal fusion, cooking, salad preparation, node localization, feature-level fusion, annotated dataset
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.