Pith. sign in

REVIEW 4 major objections 5 minor 35 references

Efficient Retail Video Annotation: A Robust Key Frame Generation Approach for Product and Customer Interaction Analysis

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Off-the-shelf YOLO detection can auto-annotate retail video, leaving under 5% of footage to humans.

desk verdict The central '95% auto-annotation' claim is internally contradicted by the paper's own IOU table; a serious referee would catch it, but the underlying workflow is worth a corrected look. read the letter →

arxiv 2506.14854 v2 pith:XAPNSQPL submitted 2025-06-17 cs.CV cs.AIcs.HCcs.LG

Varun Mannam , Zhenyu Shi This is my paper · ORCID
classification cs.CVcs.AIcs.HCcs.LG
keywords retailvideoannotationkeyframegenerationobjectdetectionYOLOcostreductionboundingboxinterpolationcustomer-productinteractionlabeling
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

The paper sets out to show that retail video annotation can be automated by using object-detection confidence to decide which frames matter. Its key-frame generation (KFG) pipeline runs a YOLO (You Only Look Once) detector on every frame, treats high-confidence detections as final annotations, sends mid-confidence frames to a human for verification, and fills low-confidence frames by interpolation. On 330 retail videos containing at least one customer, the paper reports that about 95% of videos (316 of 330) were auto-annotated with intersection-over-union above 0.5 and no human verification, leaving roughly 4-5% of videos for traditional labeling. The payoff it claims is at least a 2-fold annotation cost saving, with a potential 20-fold saving before quality checks, which matters because manual retail annotation is expensive and sensitive to arbitrary frame selection.

What carries the argument

The load-bearing machinery is the KFG thresholding pipeline. A YOLOv5x or YOLOv8x detector is run over every frame, and each person detection is bucketed by confidence: high-confidence frames (e.g., $p_{\mathrm{det}} > 0.5$) become final annotations, mid-confidence frames go to a human for verify-and-edit, and low-confidence frames are skipped and later filled by linear interpolation between neighboring key-frame boxes. Mean intersection-over-union (IOU) against human-annotated ground truth is the evaluation metric, and the fraction of videos whose high-confidence frames cover the whole clip is the source of the cost-saving figures.

What would settle it

Take the 316 videos the KFG method auto-annotated at threshold 0.5, have human annotators label every frame, and compute mean IOU per frame between the KFG/interpolated boxes and those dense labels. If mean per-frame IOU on the auto-annotated videos drops below 0.5, or if interpolated frames show systematic errors that the high-confidence key-frames do not, the claim that these videos need no human verification is refuted.

Watch

Extended reading notes

Core claim

The central claim is that confidence-based frame selection turns an off-the-shelf detector into a complete annotation pipeline. With the detection threshold set at 0.5, YOLOv5x automatically annotated 316 of 330 retail videos with IOU exceeding 0.5, and only 14 videos still required the full human-annotation workflow; YOLOv8x left 16 videos requiring human attention. The paper also reports that at threshold 0.8, 284 of 330 videos needed no human verification, and that the per-video IOU distribution at threshold 0.5 puts 155 videos above IOU 0.5. In the authors' reading, these numbers establish that KFG meets human-level annotation quality while cutting cost and average annotation time.

Load-bearing premise

The load-bearing premise is that high-confidence YOLO detections can be used as final annotations and that linear interpolation between them preserves quality; the paper provides no per-frame validation of interpolated boxes or detector accuracy against human labels, so the quality of the auto-annotated frames is unmeasured.

Editorial extensions

If this is right

  • At threshold 0.5, only 14 of 330 retail videos (YOLOv5x) or 16 (YOLOv8x) need the traditional full annotation process, giving a roughly 20-fold cost reduction before quality checks.
  • Raising the detection threshold reduces the number of frames that need human review, but the paper's test-video table shows IOU can drop to 0.29 when sparse detection leaves only 4 frames to interpolate across a 335-frame video.
  • The interpolation step lets the generated boxes feed downstream retail analytics, such as shopper journey analysis and product interaction detection, without per-frame human labels.
  • KFG can be combined with embedding-and-clustering key-frame selection, with clustering identifying behavioral phases and detector boxes supplying the object-level annotations those phases need.

Reading between the lines

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

  • The headline 95% automation counts videos, not frames: the paper's own Table 2 shows median auto-detected frame rates around 18%, so most frames in an 'automated' video are interpolation fill-ins; per-frame quality is the unmeasured quantity that really decides whether the cost savings survive.
  • A natural calibration rule follows from the paper's threshold trade-off: sweep $0.5$ to $0.8$ per store or per time-of-day and pick the threshold that maximizes mean IOU per annotation dollar, which is a testable extension the paper does not run.
  • Transferring the same confidence-threshold machinery to product or interaction labeling would require a detector with comparable precision for those classes; the paper only demonstrates person bounding boxes, so the generality claim is unsupported.
  • Validation on an open retail-like benchmark rather than a commercial dataset with a single annotation standard would clarify how much of the 95% figure is method versus dataset-specific.
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 / 5 minor

Summary. The paper proposes a Key Frame Generation (KFG) method for retail video annotation: run off-the-shelf YOLOv5x/YOLOv8x detectors on video frames, divide detections by confidence thresholds into auto-accepted, human-verified, and interpolated frames, and compare the resulting annotations with human ground truth using mean IOU. The evaluation is performed on 330 retail videos containing at least one person. The paper claims that KFG automatically annotates about 95% of videos with IOU exceeding 0.5, requires human verification on less than 5% of frames, achieves accuracy comparable to human annotators, and yields about 2x cost savings.

Significance. If the headline claims were supported, the paper would describe a valuable practical tool for reducing retail video annotation cost. The proposed pipeline is simple and easy to deploy, and the authors report experiments on a substantial commercial dataset, including a concrete cost calculation in Appendix A. However, the central quantitative claim is internally contradicted by the paper's own Table 4, and the evaluation lacks a human-human baseline, error bars, and validation of interpolated boxes. With the corrected 155-out-of-330 figure, the contribution reduces to an annotation-assistance tool with a modest mean IOU of 0.51, which may still be useful but does not support the paper's stated claims of eliminating human verification or doubling cost savings.

major comments (4)
  1. [Section 4.3.3, Table 4] The claim that "Our proposed KFG method delivered automatic annotations for approximately 95% of retail videos (316 out of 330) with an IOU metric exceeding 0.5" is directly contradicted by Table 4, which reports only 155 videos (YOLOv5x) and 156 videos (YOLOv8x) with IOU above 0.5. The 316 figure in Table 2 is the number of videos with at least one detection above the confidence threshold th1=0.5, not the number of videos whose full-video annotations satisfy the IOU quality threshold. The abstract, Table 3, and the conclusion repeat the 95% and less-than-5%-human-verification claims, so the central reported result is unsupported by the paper's own data and must be corrected.
  2. [Section 4.2, Section 4.3.3] The claim of "accuracy comparable to human annotator labeling" is not established because no human-human IOU baseline is measured. The reported mean IOU of 0.51 for YOLOv5x is modest, and without knowing the human-human agreement on the same videos, the comparison to human annotation quality is not interpretable. Additionally, Section 5 concludes that "less than 5% of frames" require human annotation, but Table 3 lists KFG annotated frames as 17.95% and human verification as 4.25% of videos, not frames; this conflation of frames and videos is a load-bearing error in the cost-savings argument.
  3. [Section 4.3.3, Table 3] The cost-savings claims are internally inconsistent. Table 3 reports that only 4.25% of videos need human annotations, which would suggest roughly 20x savings, while the text later states that 46.96% of videos (155/330) can be automatically annotated with IOU exceeding 0.5, "translating to annotation cost savings of at least 2×." These two numbers cannot both describe the same operating point, and no explicit cost model connecting the IOU threshold to annotation cost is provided. The paper should clearly state the relationship between the confidence threshold, the IOU threshold, the fraction of auto-annotated videos, and the claimed cost savings.
  4. [Section 3, Figure 6] The quality of automatically annotated frames is not directly measured. Table 2 counts frames with detections above a confidence threshold, but Table 4 reports IOU on full-video annotations "including interpolation" (Figure 6 caption). The paper never validates the interpolated bounding boxes against human labels, nor does it report per-frame YOLO accuracy against human annotations. Since the central claim is that the remaining frames can be automated without reducing annotation quality, the evaluation must include a per-frame or segment-level comparison and should report interpolation accuracy separately from detection accuracy.
minor comments (5)
  1. [Abstract] The phrase "achieving accuracy comparable to human annotator labeling" should be replaced with the measured IOU values, since no human-human baseline is provided.
  2. [Section 4.3.1, Section 4.3.2] The FFMPEG and K-means baselines do not produce bounding boxes, so their "cost savings" numbers are not directly comparable to KFG, which does produce boxes; this should be stated explicitly when comparing cost savings.
  3. [Appendix Table 5] The test video example shows that increasing the detection threshold to 0.8 reduces the number of detected frames to 4 and lowers mean IOU to 0.29, which illustrates the threshold trade-off but is only briefly discussed in the main text; a fuller discussion would help the reader interpret Table 2.
  4. [Section 4.1] The dataset description reports 935 videos, but the experiments use 330 videos containing at least one person; the filtering criteria and the number of excluded videos should be stated explicitly.
  5. [Table 3] The column "Human annotated frames (%)" is listed as NA for KFG rows, but this number is needed to support the claim about the fraction of frames requiring human verification.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the evaluation is external, comparing YOLO detections against human-annotated ground truth with IOU.

full rationale

The KFG method selects frames by a detection-probability threshold on off-the-shelf YOLO models and compares the resulting bounding boxes with professionally human-annotated ground truth using IOU (Section 4.2, Figure 4). No parameter in the derivation is fitted to the target result: thresholds are swept from 0.2 to 0.9 and reported as a trade-off table (Table 2), and the quality metric (IOU) is computed against independent human labels rather than being defined to match them. The claims of cost savings and reduced human verification follow from the number of frames that pass the detection threshold, which is a definitional property of the method, but the paper also independently evaluates annotation quality via IOU against human annotations. There are no load-bearing self-citations: the method relies on external, standard object detectors (YOLOv5x/YOLOv8x) and standard IOU metrics. The internal contradiction between the '316 out of 330 with IOU exceeding 0.5' statement in Section 4.3.3 and the 155/330 count in Table 4 is a real correctness/reporting concern, but it is not a circularity reduction: the claimed result is contradicted by the paper's own external evaluation rather than being equivalent to its input by construction. Therefore the circularity score is 0.

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

The central elements are an off-the-shelf detector and two confidence thresholds. The thresholds are fitted to the data to produce the reported cost savings. No new entities are introduced.

free parameters (4)
  • Detection probability threshold th1 = 0.5
    Frames with detection probability above th1 are auto-annotated without human review; the value is presented as an example but used in the headline results and appears tuned to balance frame yield and IOU.
  • Low-confidence threshold th2 = 0.3
    Frames below th2 are skipped and filled via interpolation; the value is arbitrary and not justified.
  • IOU success threshold = 0.5
    Videos with mean IOU above this value are counted as having acceptable automated annotations; the threshold is a common convention but not validated for retail tasks.
  • K-means cluster count K (baseline) = 10
    For the comparison deep-learning method, K was set to 10 via the Elbow method; this is a free parameter of the baseline, not the main method.
assumptions (4)
  • domain assumption Off-the-shelf YOLOv5x and YOLOv8x generalize from COCO to retail scenes without fine-tuning.
    Section 3 uses YOLO models as-is, with no adaptation to retail data.
  • domain assumption Linear or cubic-spline interpolation between key frames produces accurate bounding boxes for skipped frames.
    Sections 3 and 4.3.3 rely on interpolation for low-confidence frames, but its accuracy is never validated.
  • domain assumption Human annotations in the commercial dataset are ground truth and uniformly accurate.
    Section 4.1 states two annotators plus auditor verification, but no inter-annotator agreement is reported.
  • domain assumption Mean IOU is a sufficient quality metric for retail annotation.
    Section 4.2 declares IOU the standard measure; other error modes such as class confusion or localization consistency are ignored.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient Retail Video Annotation: A Robust Key Frame Generation Approach for Product and Customer Interaction Analysis." pith.science (2026). https://pith.science/paper/XAPNSQPL

@misc{pith2026250614854,
  author       = {Pith},
  title        = {Pith review of: Efficient Retail Video Annotation: A Robust Key Frame Generation Approach for Product and Customer Interaction Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XAPNSQPL}},
  note         = {Machine review of arXiv:2506.14854}
}
read the original abstract

Accurate video annotation plays a vital role in modern retail applications, including customer behavior analysis, product interaction detection, and in-store activity recognition. However, conventional annotation methods heavily rely on time-consuming manual labeling by human annotators, introducing non-robust frame selection and increasing operational costs. To address these challenges in the retail domain, we propose a deep learning-based approach that automates key-frame identification in retail videos and provides automatic annotations of products and customers. Our method leverages deep neural networks to learn discriminative features by embedding video frames and incorporating object detection-based techniques tailored for retail environments. Experimental results showcase the superiority of our approach over traditional methods, achieving accuracy comparable to human annotator labeling while enhancing the overall efficiency of retail video annotation. Remarkably, our approach leads to an average of 2 times cost savings in video annotation. By allowing human annotators to verify/adjust less than 5% of detected frames in the video dataset, while automating the annotation process for the remaining frames without reducing annotation quality, retailers can significantly reduce operational costs. The automation of key-frame detection enables substantial time and effort savings in retail video labeling tasks, proving highly valuable for diverse retail applications such as shopper journey analysis, product interaction detection, and in-store security monitoring.

Figures

Figures reproduced from arXiv: 2506.14854 by the authors.

Figure 1
Figure 1. Overview of the current video annotation system. To reduce the Average Annotation Time (AAT), the human [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Comparison of key frame generation methods. (a) traditional approach: The raw video is down-sampled [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Annotation of the ground truth key frames in the Video dataset (includes person, animal and vehicle classes). [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Intersection-over-Union (IOU) metrics calculation using the human annotator selective labeled frames and the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Illustration of key-frames annotation rate using FFMPEG addon and compare with the human annotations [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Illustration of videos count vs. calculated IOU when using the key-frames including interpolation in the KFG [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 24 canonical work pages

  1. [1]

    Imagenet large scale visual recognition challenge.International journal of computer vision, 115:211–252, 2015

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al. Imagenet large scale visual recognition challenge.International journal of computer vision, 115:211–252, 2015

    2015

  2. [2]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. Microsoft coco: Common objects in context. InComputer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13, pages 740–755. Springer, 2014

    2014

  3. [3]

    Simple online and realtime tracking with a deep association metric

    Nicolai Wojke, Alex Bewley, and Dietrich Paulus. Simple online and realtime tracking with a deep association metric. In2017 IEEE international conference on image processing (ICIP), pages 3645–3649. IEEE, 2017

    2017

  4. [4]

    The pascal visual object classes challenge: A retrospective.International journal of computer vision, 111:98–136, 2015

    Mark Everingham, SM Ali Eslami, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. The pascal visual object classes challenge: A retrospective.International journal of computer vision, 111:98–136, 2015

    work page 2015

  5. [5]

    Video summarization using deep semantic features

    Mayu Otani, Yuta Nakashima, Esa Rahtu, Janne Heikkilä, and Naokazu Yokoya. Video summarization using deep semantic features. InComputer Vision–ACCV 2016: 13th Asian Conference on Computer Vision, Taipei, Taiwan, November 20-24, 2016, Revised Selected Papers, Part V 13, pages 361–377. Springer, 2017

    work page 2016

  6. [6]

    Drop an octave: Reducing spatial redundancy in convolutional neural networks with octave convolution

    Yunpeng Chen, Haoqi Fan, Bing Xu, Zhicheng Yan, Yannis Kalantidis, Marcus Rohrbach, Shuicheng Yan, and Jiashi Feng. Drop an octave: Reducing spatial redundancy in convolutional neural networks with octave convolution. InProceedings of the IEEE/CVF international conference on computer vision, pages 3435–3444, 2019

    2019

  7. [7]

    Vivit: A video vision transformer

    Anurag Arnab, Mostafa Dehghani, Georg Heigold, Chen Sun, Mario Lu ˇci´c, and Cordelia Schmid. Vivit: A video vision transformer. InProceedings of the IEEE/CVF international conference on computer vision, pages 6836–6846, 2021

    2021

  8. [8]

    Time-contrastive networks: Self-supervised learning from video

    Pierre Sermanet, Corey Lynch, Yevgen Chebotar, Jasmine Hsu, Eric Jang, Stefan Schaal, Sergey Levine, and Google Brain. Time-contrastive networks: Self-supervised learning from video. In2018 IEEE international conference on robotics and automation (ICRA), pages 1134–1141. IEEE, 2018. 10 Key Frame Generation using Object DetectionA PREPRINT

    work page 2018

Show all 35 references
  1. [9]

    Accurate 3d face reconstruction with weakly-supervised learning: From single image to image set

    Yu Deng, Jiaolong Yang, Sicheng Xu, Dong Chen, Yunde Jia, and Xin Tong. Accurate 3d face reconstruction with weakly-supervised learning: From single image to image set. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, pages 0–0, 2019

    2019

  2. [10]

    Training data-efficient image transformers & distillation through attention

    Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, and Hervé Jégou. Training data-efficient image transformers & distillation through attention. InInternational conference on machine learning, pages 10347–10357. PMLR, 2021

    2021

  3. [11]

    Large scale fine-grained categorization and domain-specific transfer learning

    Yin Cui, Yang Song, Chen Sun, Andrew Howard, and Serge Belongie. Large scale fine-grained categorization and domain-specific transfer learning. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4109–4118, 2018

    2018

  4. [12]

    Rethinking the hyperparameters for fine-tuning.arXiv preprint arXiv:2002.11770, 2020

    Hao Li, Pratik Chaudhari, Hao Yang, Michael Lam, Avinash Ravichandran, Rahul Bhotika, and Stefano Soatto. Rethinking the hyperparameters for fine-tuning.arXiv preprint arXiv:2002.11770, 2020

    2002 arXiv

  5. [13]

    Model-agnostic meta-learning for fast adaptation of deep networks

    Chelsea Finn, Pieter Abbeel, and Sergey Levine. Model-agnostic meta-learning for fast adaptation of deep networks. InInternational conference on machine learning, pages 1126–1135. PMLR, 2017

    2017

  6. [14]

    Generalizing from a few examples: A survey on few-shot learning.ACM computing surveys (csur), 53(3):1–34, 2020

    Yaqing Wang, Quanming Yao, James T Kwok, and Lionel M Ni. Generalizing from a few examples: A survey on few-shot learning.ACM computing surveys (csur), 53(3):1–34, 2020

    2020

  7. [15]

    Trackingnet: A large-scale dataset and benchmark for object tracking in the wild

    Matthias Muller, Adel Bibi, Silvio Giancola, Salman Alsubaihi, and Bernard Ghanem. Trackingnet: A large-scale dataset and benchmark for object tracking in the wild. InProceedings of the European conference on computer vision (ECCV), pages 300–317, 2018

    2018

  8. [16]

    Video annotation and tracking with active learning.Advances in Neural Information Processing Systems, 24, 2011

    Carl V ondrick and Deva Ramanan. Video annotation and tracking with active learning.Advances in Neural Information Processing Systems, 24, 2011

    2011

  9. [17]

    Pathtrack: Fast trajectory annotation with path supervision

    Santiago Manen, Michael Gygli, Dengxin Dai, and Luc Van Gool. Pathtrack: Fast trajectory annotation with path supervision. InProceedings of the IEEE International Conference on Computer Vision, pages 290–299, 2017

    2017

  10. [18]

    A novel key-frames selection framework for comprehensive video summariza- tion.IEEE Transactions on Circuits and Systems for Video Technology, 30(2):577–589, 2019

    Cheng Huang and Hongmei Wang. A novel key-frames selection framework for comprehensive video summariza- tion.IEEE Transactions on Circuits and Systems for Video Technology, 30(2):577–589, 2019

    2019

  11. [19]

    A user attention model for video summarization

    Yu-Fei Ma, Lie Lu, Hong-Jiang Zhang, and Mingjing Li. A user attention model for video summarization. In Proceedings of the tenth ACM international conference on Multimedia, pages 533–542, 2002

    2002

  12. [20]

    Deep learning approach to key frame detection in human action videos.Recent Trends in Computational Intelligence, 1:1–17, 2020

    Ujwalla Gawande, Kamal Hajari, and Yogesh Golhar. Deep learning approach to key frame detection in human action videos.Recent Trends in Computational Intelligence, 1:1–17, 2020

    2020

  13. [21]

    Real-time keyframe extraction towards video content identifica- tion

    Maria Chatzigiorgaki and Athanassios N Skodras. Real-time keyframe extraction towards video content identifica- tion. In2009 16th International conference on digital signal processing, pages 1–6. IEEE, 2009

    2009

  14. [22]

    Cnn based key frame extraction for face in video recognition

    Xuan Qi, Chen Liu, and Stephanie Schuckers. Cnn based key frame extraction for face in video recognition. In 2018 IEEE 4th international conference on identity, security, and behavior analysis (ISBA), pages 1–8. IEEE, 2018

    2018

  15. [23]

    Keyframe-based video summarization with human in the loop

    Antti E Ainasoja, Antti Hietanen, Jukka Lankinen, and Joni-Kristian Kämäräinen. Keyframe-based video summarization with human in the loop. InVISIGRAPP (4: VISAPP), pages 287–296, 2018

    2018

  16. [24]

    An efficient keyframes selection based framework for video captioning

    Alok Singh, Loitongbam Sanayai Meetei, Salam Michael Singh, Thoudam Doren Singh, and Sivaji Bandy- opadhyay. An efficient keyframes selection based framework for video captioning. InProceedings of the 18th International Conference on Natural Language Processing (ICON), pages 2...

    2021

  17. [25]

    Self-supervised learning to detect key frames in videos.Sensors, 20(23):6941, 2020

    Xiang Yan, Syed Zulqarnain Gilani, Mingtao Feng, Liang Zhang, Hanlin Qin, and Ajmal Mian. Self-supervised learning to detect key frames in videos.Sensors, 20(23):6941, 2020

    2020

  18. [26]

    Deep unsupervised key frame extraction for efficient video classification.ACM Transactions on Multimedia Computing, Communications and Applications, 19(3):1–17, 2023

    Hao Tang, Lei Ding, Songsong Wu, Bin Ren, Nicu Sebe, and Paolo Rota. Deep unsupervised key frame extraction for efficient video classification.ACM Transactions on Multimedia Computing, Communications and Applications, 19(3):1–17, 2023

    2023

  19. [27]

    Unsupervised video summarization framework using keyframe extraction and video skimming

    Shruti Jadon and Mahmood Jasim. Unsupervised video summarization framework using keyframe extraction and video skimming. In2020 IEEE 5th International Conference on computing communication and automation (ICCCA), pages 140–145. IEEE, 2020

    2020

  20. [28]

    Interested keyframe extraction of commodity video based on adaptive clustering annotation.Applied Sciences, 12(3):1502, 2022

    Guangyi Man and Xiaoyan Sun. Interested keyframe extraction of commodity video based on adaptive clustering annotation.Applied Sciences, 12(3):1502, 2022

    2022

  21. [29]

    Summarizing videos with attention

    Jiri Fajtl, Hajar Sadeghi Sokeh, Vasileios Argyriou, Dorothy Monekosso, and Paolo Remagnino. Summarizing videos with attention. InComputer Vision–ACCV 2018 Workshops: 14th Asian Conference on Computer Vision, Perth, Australia, December 2–6, 2018, Revised Selected Papers 14, pa...

    2018

  22. [30]

    Online learnable keyframe extraction in videos and its application with semantic word vector in action recognition.Pattern Recognition, 122:108273, 2022

    GM Mashrur E Elahi and Yee-Hong Yang. Online learnable keyframe extraction in videos and its application with semantic word vector in action recognition.Pattern Recognition, 122:108273, 2022

    2022

  23. [31]

    A review of research on object detection based on deep learning

    Jun Deng, Xiaojing Xuan, Weifeng Wang, Zhao Li, Hanwen Yao, and Zhiqiang Wang. A review of research on object detection based on deep learning. InJournal of Physics: Conference Series, volume 1684, page 012028. IOP Publishing, 2020

    2020

  24. [32]

    Rethinking the inception architecture for computer vision

    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jon Shlens, and Zbigniew Wojna. Rethinking the inception architecture for computer vision. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 2818–2826, 2016

    2016

  25. [33]

    Segment anything.arXiv preprint arXiv:2304.02643, 2023

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. Segment anything.arXiv preprint arXiv:2304.02643, 2023

    2023 arXiv

  26. [34]

    End-to-end object detection with transformers

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to-end object detection with transformers. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part I 16, pages 21...

    2020

  27. [35]

    Ultralytics/YOLOv5: v5.0

    Glenn, Alex Jocher, Jirka Stoken, and Borovec. Ultralytics/YOLOv5: v5.0. Pytorch and Zenodo, 2021. Appendix A: Annotation cost calculation This section shows the annotation cost per object per frames in the video annotation using the amazon web services (AWS) sagemaker ground ...

    2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.