Pith. sign in

REVIEW 4 major objections 6 minor 79 references

Action Recognition based Industrial Safety Violation Detection

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that task-aware PPE checks outperform blanket PPE scans by 23% F1 on a 109-video industrial test set.

desk verdict A plausible action-conditioned PPE pipeline and a useful industrial dataset, but the headline F1 gain is not tied to genuine action understanding and the evaluation is too thin to support the claim. read the letter →

arxiv 2412.05531 v1 pith:H5ETGL4D submitted 2024-12-07 cs.CV

Surya N Reddy , Vaibhav Kurrey , Mayank Nagar , Gagan Raj Gupta This is my paper · ORCID
classification cs.CV
keywords actionrecognitionpersonalprotectiveequipmentindustrialsafetyviolationdetectionSlowFastYOLOv9clip-levelevaluationsurveillancevideo
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that industrial safety violations are better caught by first recognizing what task a worker is doing and then checking only the PPE that task requires, rather than scanning for all PPE equipment at all times. On a private test set of 109 surveillance clips of a manufacturing floor, the activity-conditioned pipeline raises the violation-detection F1-score to 0.73, compared with 0.59 for a common-PPE baseline and 0.54 for an all-PPE baseline; the authors report this as a 23% improvement. The paper also introduces a new industrial action-recognition dataset of 2,900 clips with dense spatio-temporal annotations and a human-evaluator comparison in which the model's 0.93 recall exceeds the 0.78 recall of the majority human answer.

What carries the argument

The load-bearing object is the action-to-PPE lookup table (Table 1), which maps ten shop-floor actions to the subsets of shoes, helmet, gloves, and welding helmet required for each. Algorithm 1 chains three components around this table: a SlowFast video model supplies per-person action labels and locations, a YOLOv9 detector finds PPE instances in the first, middle, and last frames of a 15-second clip, and a compliance checker compares the detected items against the table's entry for the predicted action. A clip is flagged as a violation when at least one frame contains a person missing a required item; requiring two frames instead of one trades recall (0.93 down to 0.83) for precision (0.60 up to 0.64).

What would settle it

Give the pipeline the ground-truth action labels for the 109 test videos instead of the predicted ones and measure the clip-level F1: if the score is unchanged, the action recognizer is not the source of the 0.73 F1. Alternatively, split the test set by whether the top-3 predicted actions for a clip share the same required-PPE list, and check whether the recall gain concentrates in the group where the action label actually changes the PPE check.

Watch

Extended reading notes

Core claim

The central claim is that action recognition and PPE object detection should be composed in a specific order: classify each person's action, look up the required PPE for that action from a hand-built action-to-PPE table, and then check only those items. The authors argue that task-agnostic PPE checks create false alarms because not every item is required for every task, and they report that this ordering raises clip-level recall from 0.55 on a common-PPE baseline to 0.93 while holding precision at 0.60. The result is presented as a system, with a SlowFast action model feeding bounding-box and action information to a YOLOv9 PPE detector, followed by a compliance module that marks a clip as a violation if a violating frame is found in one or two sampled frames.

Load-bearing premise

The gains depend on the action recognizer's labels being accurate enough that the PPE requirements looked up for them match the task actually being performed; the paper does not trace how action-recognition errors affect the final violation decision.

Editorial extensions

If this is right

  • A deployed system could cut false alarms on large shop floors because a worker walking is checked only for helmet and shoes, not for gloves or a welding helmet.
  • At 1.76 seconds per 15-second clip and 25 concurrent streams on the reported hardware, the pipeline is fast enough for real-time surveillance.
  • If the 0.93 recall generalizes, automated screening would catch more violations than the human evaluators in the paper's comparison (majority-answer recall 0.78), while still leaving final decisions to safety officers.
  • The two-frame violation rule gives operators a precision-recall dial: one frame maximizes catching violations, two frames reduces false positives.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The action recognizer's mean AP@0.5IOU is only 0.1093 even though Recall@Top3 is 0.94; the pipeline may be riding on the fact that most actions share the same PPE requirements, so even a wrong action label within the top few choices yields the same required-PPE list.
  • The paper does not compare the activity-conditioned system against itself with ground-truth action labels; running that ablation would isolate how much of the F1 gain is due to action understanding rather than to the rule-based per-action PPE policy.
  • Because the dataset is class-imbalanced for PPE (5,289 safety-helmet instances versus 135 welding-helmet instances), the headline 23% may change if rare-PPE actions are weighted differently or augmented.
  • The one-frame violation rule defines a clip as a violation if any sampled frame has any missing item; a deployment tuning for fewer false alarms could adopt the two-frame rule at only a 0.10 recall cost.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes an industrial safety violation detection system that first applies a SlowFast action recognition model to identify the activity of each worker in a video clip, then runs a YOLOv9 PPE detector on three sampled frames and checks only the PPE items required by the action, using a manually specified action-to-PPE mapping (Table 1). The authors introduce a new dataset of 2,900 15-second clips collected from surveillance cameras in a steel plant, report action recognition and PPE detection results, and claim a 23% improvement in F1 over PPE-only baselines on a 109-video test set. A small human study is also reported to support the claim that the system catches violations comparably to humans.

Significance. If validated, the claimed 23% F1 improvement would be practically significant for automated industrial safety monitoring, and the proposed dataset would be a useful resource for research on industrial action recognition and PPE compliance. The paper has real strengths: it collects and densely annotates a realistic industrial surveillance dataset, benchmarks multiple object detectors on it, and provides a real-time inference analysis. However, the central claim is not currently supported by the evidence: the evaluation is based on a small hand-selected test set with no statistical intervals, the connection between the weak action model and the final violation detection is not analyzed, and the ground-truth label definition may share the same action-to-PPE assumption as the algorithm. These gaps prevent the reader from attributing the reported gains to the proposed action-conditioning mechanism.

major comments (4)
  1. [5.3, Table 10] The headline 23% F1 improvement is based on a single 109-video test set (54 violations) with no confidence intervals or significance testing. The test clips were selected to ensure 'coverage of different types of actions and violations,' an ad hoc criterion, and the absolute F1 difference (0.73 vs 0.59) is within plausible noise for this sample size. Please report bootstrap confidence intervals for precision, recall, and F1, and provide a per-action or per-required-PPE breakdown of the 54 violation clips.
  2. [5.1, Tables 6-7, Algorithm 1] The action model has a mean AP@0.5IOU of only 0.1093 (Table 7), yet Algorithm 1 uses the predicted action to select the required PPE items. The paper does not state whether top-1 or top-3 predictions are used, and it does not analyze how action misclassification propagates to the clip-level violation decision. Since Table 1 shows that most actions require the same PPE set (shoes and helmet), high recall could be achieved by a checker that mostly predicts common-PPE actions, without genuine action understanding. Please add oracle-action experiments (feeding ground-truth actions instead of predicted ones), a comparison of top-1 vs top-3 inference, and an analysis of the required-PPE distribution in the 54 violation clips.
  3. [5.3, Table 1] The ground-truth violation labels for the 109-video test set appear to be defined by the same action-specific PPE requirements that the algorithm uses, but the paper never states whether the safety experts who provided ground truth worked independently of Table 1. If the labels and the algorithm share the same mapping, the evaluation is circular with respect to the proposed mechanism. Please specify the annotation protocol for the test set, including whether the experts had access to Table 1, or re-annotate the test set with independent experts.
  4. [4.3, Algorithm 1] The clip-level violation metric samples only three frames (F_first, F_middle, F_last) of each 15-second clip, while the earlier text mentions 15 frames; the assumption that these three frames capture all safety-relevant information is unjustified and could bias the reported recall. Please evaluate the sensitivity of the clip-level results to the number and positions of sampled frames, or provide evidence that violations are persistent enough to be detected in the three selected frames.
minor comments (6)
  1. [4.3] The prose says 'even if 1 frame of the 15 clips has any PPE violation detected' but the formula and Algorithm 1 use N=3 frames; please reconcile the notation.
  2. [5.3] The human study reports an 'average recall of 0.78%,' which is presumably a typo for 78% or 0.78; also specify whether the 20-video human evaluation set overlaps with the 109-video test set.
  3. [3.1.2, Table 3] The action taxonomy in the text lists seven coarse actions, while Table 3 lists twelve micro-action classes; please clarify the mapping between the two levels.
  4. [Tables 2 and 5] Table 2 reports 'Our Data' with 7 classes and 3,000 images for PPE detection, while the action dataset has 12 classes and 2,900 clips; make explicit that these are different annotation layers of the same collected footage.
  5. [Algorithm 1] The algorithm uses symbols B_info, BPPE, and PPE_List without definitions; please define them in the text or in the caption.
  6. [Dataset availability] The paper proposes a novel dataset but provides no URL, repository, or availability statement; as a dataset contribution, a public release or at least a detailed availability plan is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No demonstrated circularity: the activity-conditioned check is an empirical composition of independently trained models, and the action-PPE table is an external domain rule rather than a fitted or self-cited input.

full rationale

Walking the derivation chain, the central claim is an empirical F1 comparison among three clip-level checkers (common PPE, all PPE, and activity-based PPE) on 109 test videos. The activity-based checker in Algorithm 1 combines two independently trained components: a SlowFast action recognition model, whose labels are defined by Table 3 micro-action definitions (e.g., welding requires observing a welding spark, not PPE presence), and a YOLOv9 PPE detector trained on the PPE classes in Table 5. The action-to-PPE mapping in Table 1 is stated as an external industrial-safety rule and is not fitted to the violation labels; the baselines use the same PPE detector and the same clip-level rule, so the reported 23% F1 gain is a measured consequence of checking only the PPE designated for the predicted action. There are no load-bearing self-citations: SlowFast, YOLOv9, AVA-style annotation, and PySlowFast are all external tools, and no author-derived uniqueness or ansatz is invoked. The low mean action AP and the absence of an error-propagation analysis are genuine interpretability and robustness limitations, but they do not make the reported improvement definitionally equal to an input. The only potential circularity would require the 109-clip ground-truth violation labels to be generated by the same Table 1/Algorithm 1 rule; the paper does not state this, so that reduction cannot be quoted or exhibited, and I do not treat an unstated labeling procedure as demonstrated circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central result depends on several hand-chosen design decisions (the action-to-PPE table, the sampling of 3 frames per clip, the test-set filtering, and the frame-count threshold), plus unvalidated domain assumptions about the correctness of the taxonomy and the generalization of models trained on a single plant's cameras.

free parameters (4)
  • Action-specific PPE requirement mapping (Table 1) = Hand-authored mapping for 10 actions to PPE sets
    This mapping determines which PPE items are checked for each predicted action; it is a human choice, not learned, and it also appears to define what counts as a violation in the test ground truth.
  • Violation flag threshold (number of frames) = 1 frame for the headline result, 2 frames for an alternative
    The paper evaluates both rules and reports the 1-frame rule as primary because it maximizes recall (Section 5.3); this is post-hoc selection on the test set.
  • Frames sampled per clip for PPE check = 3 frames (first, middle, last) out of a 15-second, 180-frame clip
    This sampling choice controls which moments can be flagged as violations and can miss transient PPE removal.
  • Test set selection criteria = 109 clips with sufficient field of view and coverage of actions/violations
    The test clips are filtered by FOV and action/violation coverage, which is a manual selection and may not represent the full deployment distribution.
assumptions (4)
  • domain assumption The 12-category action taxonomy and Table 1 PPE requirements match real safety standards in the target plant.
    The system's correctness depends on this mapping being authoritative; it is authored by the researchers with input from the plant context, but not independently validated or released.
  • domain assumption The SlowFast and YOLOv9 models generalize from the annotated training clips to the broader surveillance feed distribution.
    All experiments are on data from the same three cameras and same plant; no cross-site or external validation is provided.
  • domain assumption Crowdsourced annotations of actions and bounding boxes are accurate enough for training and evaluation.
    No inter-annotator agreement metrics or quality control statistics are reported for the 45,652 action instances.
  • ad hoc to paper The 3-frame sampling rule preserves all safety-relevant information in a 15-second clip.
    The paper chooses first/middle/last frames without analysis of how much violation time is missed between the sampled frames.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Action Recognition based Industrial Safety Violation Detection." pith.science (2026). https://pith.science/paper/H5ETGL4D

@misc{pith2026241205531,
  author       = {Pith},
  title        = {Pith review of: Action Recognition based Industrial Safety Violation Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H5ETGL4D}},
  note         = {Machine review of arXiv:2412.05531}
}
read the original abstract

Proper use of personal protective equipment (PPE) can save the lives of industry workers and it is a widely used application of computer vision in the large manufacturing industries. However, most of the applications deployed generate a lot of false alarms (violations) because they tend to generalize the requirements of PPE across the industry and tasks. The key to resolving this issue is to understand the action being performed by the worker and customize the inference for the specific PPE requirements of that action. In this paper, we propose a system that employs activity recognition models to first understand the action being performed and then use object detection techniques to check for violations. This leads to a 23% improvement in the F1-score compared to the PPE-based approach on our test dataset of 109 videos.

Figures

Figures reproduced from arXiv: 2412.05531 by the authors.

Figure 1
Figure 1. Sample Data with Multi actor- Multi Action Indus [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Distribution of Action Labels in Proposed Dataset [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Combined action recognition and PPE detection for [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

79 extracted references · 67 canonical work pages

  1. [1]

    Sami Abu-El-Haija, Nisarg Kothari, Joonseok Lee, Apostol Natsev, George Toderici, Balakrishnan Varadarajan, and Sudheendra Vijayanarasimhan. 2016. YouTube-8M: A Large-Scale Video Classification Benchmark. ArXiv (2016)

    work page 2016

  2. [2]

    Hafiz Mughees Ahmad and Afshin Rahimi. 2024. SH17: A Dataset for Human Safety and Personal Protective Equipment Detection in Manufacturing Industry

    work page 2024

  3. [3]

    Gedas Bertasius, Heng Wang, and Lorenzo Torresani. 2021. Is Space-Time Atten- tion All You Need for Video Understanding? ArXiv (2021)

    work page 2021

  4. [4]

    Moshe Blank, Lena Gorelick, Eli Shechtman, Michal Irani, and Ronen Basri. 2005. Actions as space-time shapes. Tenth IEEE International Conference on Computer Vision (ICCV’05) Volume 1 (2005)

    work page 2005

  5. [5]

    Zhe Cao, Tomas Simon, Shih-En Wei, and Yaser Sheikh. 2016. Realtime Multi- person 2D Pose Estimation Using Part Affinity Fields. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2016), 1302–1310

    work page 2016

  6. [6]

    João Carreira and Andrew Zisserman. 2017. Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017), 4724–4733. https://api.semanticscholar. org/CorpusID:206596127

    work page 2017

  7. [7]

    Shi Chen and Kazuyuki Demachi. 2020. A Vision-Based Approach for Ensuring Proper Use of Personal Protective Equipment (PPE) in Decommissioning of Fukushima Daiichi Nuclear Power Station. Applied Sciences (2020)

    work page 2020

  8. [8]

    Mejdi DALLEL, Vincent HAVARD, David BAUDRY, and Xavier SAVATIER. 2020. InHARD - Industrial Human Action Recognition Dataset in the Context of In- dustrial Collaborative Robotics. In 2020 IEEE International Conference on Human- Machine Systems (ICHMS)

    work page 2020

Show all 79 references
  1. [9]

    Mejdi Dallel, Vincent Havard, David Baudry, and Xavier Savatier. 2020. InHARD - Industrial Human Action Recognition Dataset in the Context of Industrial Collaborative Robotics. 2020 IEEE International Conference on Human-Machine Systems (ICHMS) (2020)

    2020

  2. [10]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In North American Chapter of the Association for Computational Linguistics

    2019

  3. [11]

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xi- aohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. 2020. An Image is Worth 16x16 Words: Transformers for Image Recogn...

    2020

  4. [12]

    Haoqi Fan, Yanghao Li, Bo Xiong, Wan-Yen Lo, and Christoph Feichtenhofer

  5. [13]

    Christoph Feichtenhofer. 2020. X3D: Expanding Architectures for Efficient Video Recognition. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion (CVPR) (2020)

    2020

  6. [14]

    Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik, and Kaiming He. 2018. Slow- Fast Networks for Video Recognition. 2019 IEEE/CVF International Conference on Computer Vision (ICCV) (2018)

    2018

  7. [15]

    Ross Girshick. 2015. Fast r-cnn. In Proceedings of the IEEE international conference on computer vision. 1440–1448

    2015

  8. [16]

    Ross, George Toderici, Yeqing Li, Susanna Ricco, Rahul Sukthankar, Cordelia Schmid, and Jitendra Malik

    Chunhui Gu, Chen Sun, Sudheendra Vijayanarasimhan, Caroline Pantofaru, David A. Ross, George Toderici, Yeqing Li, Susanna Ricco, Rahul Sukthankar, Cordelia Schmid, and Jitendra Malik. 2017. AVA: A Video Dataset of Spatio- Temporally Localized Atomic Visual Actions. 2018 IEEE/C...

    2017

  9. [17]

    Fabian Caba Heilbron, Victor Escorcia, Bernard Ghanem, and Juan Carlos Niebles

  10. [18]

    in the wild

    Haroon Idrees, Amir Zamir, Yu-Gang Jiang, Alexander N. Gorban, Ivan Laptev, Rahul Sukthankar, and Mubarak Shah. 2016. The THUMOS challenge on action recognition for videos "in the wild". ArXiv (2016)

    2016

  11. [19]

    Francesco Iodice, Elena De Momi, and Arash Ajoudani. 2022. HRI30: An Ac- tion Recognition Dataset for Industrial Human-Robot Interaction. In 2022 26th International Conference on Pattern Recognition (ICPR)

    2022

  12. [20]

    Francesco Iodice, Elena De Momi, and Arash Ajoudani. 2022. HRI30: An Ac- tion Recognition Dataset for Industrial Human-Robot Interaction. 2022 26th International Conference on Pattern Recognition (ICPR) (2022)

    2022

  13. [21]

    Vukicevic

    Velibor Isailović, Aleksandar Peulić, Marko Djapan, Marija Savković, and Arso M. Vukicevic. 2022. The compliance of head-mounted industrial PPE by using deep learning object detectors. Scientific Reports (2022)

    2022

  14. [22]

    Hueihan Jhuang, Juergen Gall, Silvia Zuffi, Cordelia Schmid, and Michael J. Black

  15. [23]

    Andrej Karpathy, George Toderici, Sanketh Shetty, Thomas Leung, Rahul Suk- thankar, and Li Fei-Fei. 2014. Large-Scale Video Classification with Convolutional Neural Networks. 2014 IEEE Conference on Computer Vision and Pattern Recogni- tion (2014)

    2014

  16. [24]

    Yan Ke, Rahul Sukthankar, and Martial Hebert. 2005. Efficient visual event detection using volumetric features. Tenth IEEE International Conference on Computer Vision (ICCV’05) Volume 1 (2005)

    2005

  17. [25]

    Kyunghwan Kim, Kangeun Kim, and Soyoon Jeong. 2023. Application of YOLO v5 and v8 for Recognition of Safety Risk Factors at Construction Sites.Sustainability 15 (10 2023), 15179

    2023

  18. [26]

    Yu Kong and Yun Raymond Fu. 2018. Human Action Recognition and Prediction: A Survey. International Journal of Computer Vision 130 (2018), 1366 – 1401. https://api.semanticscholar.org/CorpusID:49551723

    2018

  19. [27]

    Kuehne, H

    H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre. 2011. HMDB: a large video database for human motion recognition. In Proceedings of the International Conference on Computer Vision (ICCV)

    2011

  20. [28]

    Poggio, and Thomas Serre

    Hilde Kuehne, Hueihan Jhuang, Estíbaliz Garrote, Tomaso A. Poggio, and Thomas Serre. 2011. HMDB: A large video database for human motion recognition. 2011 International Conference on Computer Vision (2011)

    2011

  21. [29]

    Ivan Laptev. 2005. On Space-Time Interest Points. International Journal of Computer Vision 64 (2005), 107–123. https://api.semanticscholar.org/CorpusID: 2619278

    2005

  22. [30]

    Luziwei Leng, Kaiwei Che, Kaixuan Zhang, Jianguo Zhang, Qinghu Meng, Jie Cheng, Qinghai Guo, and Jianxing Liao. 2022. Differentiable hierarchical and surrogate gradient search for spiking neural networks. In Advances in Neural Information Processing Systems

    2022

  23. [31]

    Ross, João Carreira, Alexander Vostrikov, and Andrew Zisserman

    Ang Li, Meghana Thotakuri, David A. Ross, João Carreira, Alexander Vostrikov, and Andrew Zisserman. 2020. The AVA-Kinetics Localized Human Actions Video Dataset. ArXiv (2020)

    2020

  24. [32]

    Ross, João Carreira, Alexander Vostrikov, and Andrew Zisserman

    Ang Li, Meghana Thotakuri, David A. Ross, João Carreira, Alexander Vostrikov, and Andrew Zisserman. 2020. The AVA-Kinetics Localized Human Actions Video Dataset. arXiv:2005.00214 [cs.CV]

    2020 arXiv

  25. [33]

    Li, Zhengyou Zhang, and Zicheng Liu

    W. Li, Zhengyou Zhang, and Zicheng Liu. 2010. Action recognition based on a bag of 3D points. 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops (2010)

    2010

  26. [34]

    Yixuan Li, Lei Chen, Runyu He, Zhenzhi Wang, Gangshan Wu, and Limin Wang

  27. [35]

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. 2017. Focal loss for dense object detection. In Proceedings of the IEEE international conference on computer vision . 2980–2988

    2017

  28. [36]

    Jingen Liu, Benjamin Kuipers, and Silvio Savarese. 2011. Recognizing human actions by attributes. CVPR 2011 (2011), 3337–3344. https://api.semanticscholar. org/CorpusID:9119671

    2011

  29. [37]

    Liu, Dragomir Anguelov, D

    W. Liu, Dragomir Anguelov, D. Erhan, Christian Szegedy, Scott E. Reed, Cheng- Yang Fu, and Alexander C. Berg. 2015. SSD: Single Shot MultiBox Detector. In European Conference on Computer Vision

    2015

  30. [38]

    Ze Liu, Jia Ning, Yue Cao, Yixuan Wei, Zheng Zhang, Stephen Lin, and Han Hu

  31. [39]

    Christian Mandery, Ömer Terlemez, Martin Do, Nikolaus Vahrenkamp, and Tamim Asfour. 2015. The KIT whole-body human motion database. In 2015 International Conference on Advanced Robotics (ICAR)

    2015

  32. [40]

    Marcin Marszalek, Ivan Laptev, and Cordelia Schmid. 2009. Actions in context. 2009 IEEE Conference on Computer Vision and Pattern Recognition (2009)

    2009

  33. [41]

    Naval Kishore Mehta, Shyam Sunder Prasad, Sumeet Saurav, Ravi Saini, and Sanjay Singh. 2024. IAR-Net: A Human–Object Context Guided Action Recog- nition Network for Industrial Environment Monitoring. IEEE Transactions on Instrumentation and Measurement 73 (2024), 1–8

    2024

  34. [42]

    Louis-Philippe Morency, Ariadna Quattoni, and Trevor Darrell. 2007. Latent- Dynamic Discriminative Models for Continuous Gesture Recognition. 2007 IEEE Conference on Computer Vision and Pattern Recognition (2007), 1–8. https: //api.semanticscholar.org/CorpusID:7117722

    2007

  35. [43]

    Müller, T

    M. Müller, T. Röder, M. Clausen, B. Eberhardt, B. Krüger, and A. Weber. 2007. Documentation Mocap Database HDM05. Technical Report CG-2007-2. Universität Bonn

    2007

  36. [44]

    Neftci, Hesham Mostafa, and Friedemann Zenke

    Emre O. Neftci, Hesham Mostafa, and Friedemann Zenke. 2019. Surrogate Gradi- ent Learning in Spiking Neural Networks: Bringing the Power of Gradient-based optimization to spiking neural networks. IEEE Signal Processing Magazine 36 (2019), 51–63

    2019

  37. [45]

    njvisionpower. 2024. Safety-Helmet-Wearing-Dataset. GitHub (2024)

    2024

  38. [46]

    Alnajjar, Luqman Ali, Tan-Hsu Tan, Jun-Wei Hsieh, and Ping-Yang Chen

    Munkh-Erdene Otgonbold, Munkhjargal Gochoo, Fady S. Alnajjar, Luqman Ali, Tan-Hsu Tan, Jun-Wei Hsieh, and Ping-Yang Chen. 2022. SHEL5K: An Extended Dataset and Benchmarking for Safety Helmet Detection. Sensors (Basel, Switzer- land) 22 (2022)

    2022

  39. [47]

    Paul Over, Jon Fiscus, Gregory Sanders, David Joy, Martial Michel, George Awad, Alan Smeaton, Wessel Kraaij, and Georges Quénot. 2014. TRECVID 2014 – An Overview of the Goals, Tasks, Data, Evaluation Mechanisms, and Metrics

    2014

  40. [48]

    Qinhan Xiao Qiankun Xiao, Junfeng Li. 2013. Human Motion Capture Data Retrieval Based on Quaternion and EMD. InInternational Conference on Intelligent Human-Machine Systems and Cybernetics

    2013

  41. [49]

    Girshick, and Jian Sun

    Shaoqing Ren, Kaiming He, Ross B. Girshick, and Jian Sun. 2015. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence 39 (2015), 1137–1149. CODS-COMAD Dec ’24, December 18–21, 2024, Jodhpu...

    2015

  42. [50]

    Rodriguez, Javed Ahmed, and Mubarak Shah

    Mikel D. Rodriguez, Javed Ahmed, and Mubarak Shah. 2008. Action MACH a spatio-temporal Maximum Average Correlation Height filter for action recogni- tion. 2008 IEEE Conference on Computer Vision and Pattern Recognition (2008)

    2008

  43. [51]

    M. S. Ryoo and L. Matthies. 2013. First-Person Activity Recognition: What Are They Doing to Me?. InIEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    2013

  44. [52]

    Paul Scovanner, Saad Ali, and Mubarak Shah. 2007. A 3-dimensional sift descriptor and its application to action recognition.Proceedings of the 15th ACM international conference on Multimedia (2007). https://api.semanticscholar.org/CorpusID: 1087061

    2007

  45. [53]

    Javen Qinfeng Shi, Li Cheng, Li Wang, and Alex Smola. 2011. Human Action Segmentation and Recognition Using Discriminative Semi-Markov Models. Inter- national Journal of Computer Vision 93 (2011), 22–32. https://api.semanticscholar. org/CorpusID:9054863

    2011

  46. [54]

    Sigurdsson, Gül Varol, X

    Gunnar A. Sigurdsson, Gül Varol, X. Wang, Ali Farhadi, Ivan Laptev, and Abhi- nav Kumar Gupta. 2016. Hollywood in Homes: Crowdsourcing Data Collection for Activity Understanding. In European Conference on Computer Vision

    2016

  47. [55]

    Khurram Soomro, Amir Zamir, and Mubarak Shah. 2012. UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild. ArXiv (2012)

    2012

  48. [56]

    Khurram Soomro, Amir Roshan Zamir, and Mubarak Shah. 2012. UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild. arXiv:1212.0402

    2012 arXiv

  49. [57]

    Moritz Tenorth, Jan Bandouch, and Michael Beetz. 2009. The TUM Kitchen Data Set of everyday manipulation activities for motion tracking and action recog- nition. 2009 IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops (2009)

    2009

  50. [58]

    Bourdev, Rob Fergus, Lorenzo Torresani, and Manohar Paluri

    Du Tran, Lubomir D. Bourdev, Rob Fergus, Lorenzo Torresani, and Manohar Paluri. 2014. Learning Spatiotemporal Features with 3D Convolutional Networks. 2015 IEEE International Conference on Computer Vision (ICCV) (2014), 4489–4497. https://api.semanticscholar.org/CorpusID:1122604

    2014

  51. [59]

    Du Tran, Heng Wang, Lorenzo Torresani, Jamie Ray, Yann LeCun, and Manohar Paluri. 2017. A Closer Look at Spatiotemporal Convolutions for Action Recog- nition. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (2017)

    2017

  52. [60]

    Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N

    Ashish Vaswani, Noam M. Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is All you Need. In Neural Information Processing Systems

    2017

  53. [61]

    Vukicevic, Marko Djapan, Velibor Isailović, Danko Z

    Arso M. Vukicevic, Marko Djapan, Velibor Isailović, Danko Z. Milasinovic, Marija Savković, and Pavle Miloćević. 2022. Generic compliance of industrial PPE by using deep learning techniques. Safety Science (2022)

    2022

  54. [62]

    Chien-Yao Wang, I-Hau Yeh, and Hong-Yuan Mark Liao. 2024. Yolov9: Learning what you want to learn using programmable gradient information.arXiv preprint arXiv:2402.13616 (2024)

    2024 arXiv

  55. [63]

    Verbeek, and Cordelia Schmid

    Heng Wang, Dan Oneaţă, Jakob J. Verbeek, and Cordelia Schmid. 2015. A Robust and Efficient Video Representation for Action Recognition. International Jour- nal of Computer Vision 119 (2015), 219 – 238. https://api.semanticscholar.org/ CorpusID:11491197

    2015

  56. [64]

    Zijian Wang, Yimin Wu, Lichao Yang, Arjun Thirunavukarasu, Colin Evison, and Yifan Zhao. 2021. Fast Personal Protective Equipment Detection for Real Construction Sites Using Deep Learning Approaches. Sensors (Basel, Switzerland) (2021)

    2021

  57. [65]

    Yuyang Wanyan, Xiaoshan Yang, Weiming Dong, and Changsheng Xu. 2024. A Comprehensive Review of Few-shot Action Recognition

    2024

  58. [66]

    Jixiu Wu, Nian Cai, Wenjie Chen, Huiheng Wang, and Guotian Wang. 2019. Automatic detection of hardhats worn by construction personnel: A deep learning approach and benchmark dataset. Automation in Construction (2019)

    2019

  59. [67]

    Yuxin Wu, Alexander Kirillov, Francisco Massa, Wan-Yen Lo, and Ross Girshick

  60. [68]

    Fan Yang. 2022. CustomAva. https://github.com/Whiffe/Custom-ava-dataset_ Custom-Spatio-Temporally-Action-Video-Dataset

    2022

  61. [69]

    Serena Yeung, Olga Russakovsky, Ning Jin, Mykhaylo Andriluka, Greg Mori, and Li Fei-Fei. 2015. Every Moment Counts: Dense Detailed Labeling of Actions in Complex Videos. International Journal of Computer Vision (2015)

    2015

  62. [70]

    Fusheng Yu, Xiaoping Wang, Jiang Li, Shaojin Wu, Jun jie Zhang, and Zhigang Zeng. 2023. Towards Complex Real-World Safety Factory Inspection: A High- Quality Dataset for Safety Clothing and Helmet Detection. ArXiv (2023)

    2023

  63. [71]

    Junsong Yuan, Zicheng Liu, and Ying Wu. 2009. Discriminative subvolume search for efficient action detection.2009 IEEE Conference on Computer Vision and Pattern Recognition (2009)

    2009

  64. [72]

    Faishal Zhafran, Endah Suryawati Ningrum, Mohamad Nasyir Tamara, and Eny Kusumawati. 2019. Computer Vision System Based for Personal Protective Equipment Detection, by Using Convolutional Neural Network.2019 International Electronics Symposium (IES) (2019)

    2019

  65. [73]

    Sijie Zhu, Taojiannan Yang, Mat’ias Mendieta, and Chen Chen. 2020. A3D: Adaptive 3D Networks for Video Action Recognition. ArXiv (2020)

    2020

  66. [2013]

    2013 IEEE International Conference on Computer Vision (2013)

    Towards Understanding Action Recognition. 2013 IEEE International Conference on Computer Vision (2013)

    2013

  67. [2015]

    2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015)

    ActivityNet: A large-scale video benchmark for human activity understand- ing. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015)

    2015

  68. [2019]

    https://github.com/facebookresearch/detectron2

    Detectron2. https://github.com/facebookresearch/detectron2

  69. [2020]

    https://github.com/facebookresearch/slowfast

    PySlowFast. https://github.com/facebookresearch/slowfast

  70. [2021]

    2021 IEEE/CVF International Conference on Computer Vision (ICCV) (2021)

    MultiSports: A Multi-Person Video Dataset of Spatio-Temporally Localized Sports Actions. 2021 IEEE/CVF International Conference on Computer Vision (ICCV) (2021)

    2021

  71. [2022]

    In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Video Swin Transformer. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . IEEE

    2022

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.