Pith. sign in

REVIEW 4 major objections 4 minor 57 references

Snap, Segment, Deploy: A Visual Data and Detection Pipeline for Wearable Industrial Assistants

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

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

Pith's one-line read The paper claims that a two-stage training strategy called Background-Agnostic Refinement (BAR) makes a lightweight on-device detector roughly three times more accurate under domain shift, with mAP@0.5:0.95 rising from 0.06 to 0.17 on the…

desk verdict Useful synthetic-data pipeline and a plausible-but-unproven BAR trick; evaluation is too thin for the deployment claim. read the letter →

arxiv 2507.21072 v1 pith:TZMVYVW5 submitted 2025-06-09 cs.HC cs.RO

Di Wen , Junwei Zheng , Ruiping Liu , Yi Xu , Kunyu Peng , Rainer Stiefelhagen This is my paper · ORCID
classification cs.HCcs.RO
keywords industrialassemblyassistancewearablecomputingon-deviceobjectdetectionsyntheticdatagenerationcopy-pasteaugmentationdomainshiftretrieval-augmentedGear8dataset
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 an industrial assembly assistant can run entirely on a mobile device, offline, privacy-preserving, and without manual annotation, if its training data is synthesized and its detector is refined in a specific way. The training data is built by capturing multi-view video of each part, extracting instance masks, and copy-pasting the masked parts onto industrial background images. The central training claim is Background-Agnostic Refinement (BAR): after training on those synthetic scenes, the detector is fine-tuned on its own high-confidence predictions cropped and placed on white canvases, which forces it to rely on object-intrinsic features rather than background context. On the Gear8 dataset, BAR raises mAP@0.5:0.95 from 0.06 to 0.17, and adding test-time augmentation raises mAP@0.5 to 0.30. A user study with nine participants reports that assisted assembly took about 219 seconds on average instead of 542 seconds, with no damaged parts under guidance.

What carries the argument

The load-bearing mechanism is the two-stage Background-Agnostic Refinement (BAR). In the first stage, a lightweight YOLO-style detector is trained on synthetically composited images: three to five masked part instances pasted onto one of twelve industrial backgrounds with random scale and rotation. In the second stage, the trained detector is re-run over its own training images, the high-confidence bounding boxes are cropped, and each crop is placed on a plain white canvas; that purified pseudo-labeled set is used for fine-tuning. This transfer from context-rich synthetic scenes to context-free white backgrounds is the operation the paper says decouples foreground object learning from background distractions and improves robustness.

What would settle it

On the Gear8 training set, fine-tune BAR using the detector's own high-confidence crops and, separately, using ground-truth white-background crops cut from the same synthetic images; if the pseudo-label version is worse on the corrupted test set, the load-bearing assumption fails. A simpler check is to measure label and IoU mismatch between pseudo-labels and the ground-truth masks that the synthetic pipeline already produces.

Watch

Extended reading notes

Core claim

The paper's central claim is that removing background context from a detector's own confident predictions, and then fine-tuning on those white-background crops, is an effective domain-shift countermeasure for lightweight detectors. The argument is that the first-stage model overfits to contextual artifacts in the synthetic copy-paste images; re-training on purified object crops forces the model to encode the parts themselves. The measured effect is a rise in mAP@0.5:0.95 from 0.06 to 0.17 on a real-world test set with ten types of corruption, with BAR plus test-time augmentation reaching mAP@0.5 of 0.30. The paper frames this as a training-only change that needs no architecture modification, no manual annotation, and no extra inference cost beyond optional test-time augmentation.

Load-bearing premise

The refinement step assumes that the first-pass detector's high-confidence predictions on the synthetic images are accurate enough to serve as pseudo-labels; if those boxes are misclassified or badly located, the white-background fine-tuning set contains mislabeled objects and can amplify errors instead of reducing them.

Editorial extensions

If this is right

  • If BAR works as claimed, a factory can train a usable part detector from smartphone videos and public background images, with no labeling effort.
  • The privacy claim holds as stated: detection, retrieval, dialogue, and speech synthesis all run on-device, so no assembly footage must leave the factory.
  • On the Gear8 data, BAR and test-time augmentation complement each other, while slicing-based inference (SAHI) hurts because the objects are small and well-centered.
  • The user-study result implies that guidance of this kind can cut average assembly time from roughly 9 minutes to under 4 minutes and prevent error-induced part damage at moderate cognitive load.

Reading between the lines

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

  • Editorial extension: BAR is a generic context-removal regularizer, so the same white-canvas fine-tuning could transfer to other lightweight detectors and to any domain where background correlation drives errors, not just industrial parts.
  • Editorial extension: the paper leaves the confidence threshold for pseudo-labeling unspecified; a threshold scan on a held-out portion of Gear8 could turn BAR into a calibrated procedure and would directly test the weak assumption.
  • Editorial extension: comparing BAR against an identically trained model whose second stage uses ground-truth white-background crops would isolate how much of the gain comes from pseudo-label noise versus from the background-agnostic prior itself.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper presents a wearable industrial assistant system that combines on-device object detection, speech recognition, retrieval-augmented generation, and text-to-speech to support semi-hands-free assembly guidance. To avoid manual annotation, the authors propose an automated synthetic data pipeline (Gear8) in which object masks extracted with SAM2 are copy-pasted onto industrial background images. They further introduce Background-Agnostic Refinement (BAR), a two-stage training strategy in which a YOLOv11n detector is first trained on the synthetic composites and then fine-tuned on white-background crops of its own high-confidence detections. Experiments report that BAR improves mAP@0.5:0.95 from 0.06 to 0.17 on a held-out real-image test set, and a nine-participant user study reports faster task completion, reduced part damage, and moderate cognitive load. The authors conclude that the system is a deployable, privacy-preserving solution for real-time industrial assistance.

Significance. If validated, the work would make three useful contributions: an annotation-free synthetic data pipeline using consumer-grade video capture, a simple architecture-agnostic refinement strategy that improves robustness to domain shift, and a fully on-device multimodal assistant evaluated under realistic constraints. The promised release of the Gear8 dataset and source code is a concrete asset for reproducibility and follow-up research. However, the central empirical claim about BAR is currently under-supported: the reported gain is not isolated from additional training time, the pseudo-labeling step is not validated against available ground truth, and the detection evaluation lacks error bars, significance testing, and comparison to existing synthetic-data or domain-robustness methods. The user study, while informative, is too small and too weakly analyzed to carry the system-level claims on its own. The strengths are real, but the evidence base needs substantial strengthening before the claims can be accepted.

major comments (4)
  1. [V-C, Table I] The central claim that BAR improves robustness is not isolated from the additional training that the two-stage schedule introduces. The baseline is trained until validation mAP50 exceeds 95.0, and BAR then fine-tunes on white-background crops; there is no control condition that continues training the baseline for the same number of additional steps or epochs on the original synthetic data or on any other control distribution. Without such a control, the reported gain from 0.06 to 0.17 in mAP@0.5:0.95 could be attributable to a longer schedule or more gradient steps rather than to the background-agnostic mechanism. This control is necessary to support the paper's central claim.
  2. [IV-A] The pseudo-label step in BAR is not validated despite the availability of exact ground truth. Because Gear8 is synthetically composed, the true instance masks and labels are known by construction, yet the method re-applies the first-stage detector to extract high-confidence predictions without specifying a confidence threshold or reporting the precision/recall of these pseudo-labels. If the pseudo-labels contain misclassifications or poorly localized boxes, the white-background refinement set is mislabeled, and the measured improvement could be an upper bound on what a correctly labeled refinement would achieve. The natural control is to run the same two-stage schedule using ground-truth masks for the white-crop refinement and to report pseudo-label accuracy.
  3. [V-C and III-D] The detection evaluation lacks statistical grounding. Table I reports single numbers with no error bars across seeds, no significance tests, and no per-corruption breakdown, even though the corruption test set is constructed by the authors with a fixed and unreferenced severity. In addition, there is no comparison to established synthetic-data methods or robustness strategies such as copy-paste augmentation alone, domain randomization, CycleGAN-based refinement, or other lightweight wearable detectors. As a result, the claim of 'consistent improvements across all metrics' is stronger than the evidence supports. The authors should report mean and variance over multiple seeds, test the difference between BAR and the baseline with a paired test, and include at least one comparison baseline from the related literature.
  4. [V-E] The user study claims of practical viability are not supported by inferential statistics. The study uses only nine participants, reports no paired significance test for the reduction in completion time from 542.3 s to 219.2 s, and acknowledges an outlier (participant 7) without quantifying its effect. The NASA-TLX score of 48.0 is compared to a published median but no dispersion is reported, and the overall recommendation score of 6.11 (SD = 1.62) indicates moderate and variable satisfaction. The text should either moderate the conclusions to descriptive observations or add appropriate statistical tests and confidence intervals.
minor comments (4)
  1. [III (Dataset Construction)] The sentence 'We construct the Gear8 dataset following the pipeline in Section IV-A' in Section V-A is incorrect; the data construction pipeline is described in Section III, while Section IV-A describes BAR. The cross-reference should be fixed.
  2. [IV-B-2] DepthAnything is cited as [8], but reference [8] is the RAG paper (Lewis et al.); the DepthAnything citation [51] should be used at this location.
  3. [V-E] The text refers to 'Table 4' for completion-time results, but the data appear in Figure 4(b); the reference should be corrected.
  4. [V-C, Table I] The row 'w/ BAR+SAHI' reports the same mAP@0.5:0.95 as the baseline (0.06), which is consistent with the text's claim that SAHI has a negative effect; the text would benefit from explicitly stating that BAR's gain is lost when combined with SAHI.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: BAR's improvement is measured on a held-out real-image test set, and no prediction reduces to a fitted input by construction.

full rationale

The paper's central empirical claim is that Background-Agnostic Refinement (BAR) raises mAP@0.5:0.95 from 0.06 to 0.17 (Section V-C, Table I). The comparison is between YOLOv11n trained on the synthetic Gear8 set and the same model after BAR fine-tuning, with both evaluated on a held-out test set of 196 real-world images plus 10 deterministic corruptions; the reported numbers are not fit from the test set or defined in terms of the test output. The BAR refinement set is built from the detector's own high-confidence predictions on training images, which is a self-training methodology rather than a circular derivation: the gain is an empirical outcome on held-out data and is not mathematically forced by the construction of the pseudo-labels. The paper does not calibrate its corruption severities to external measurements and does not include a control that trains the baseline for the same number of extra steps on the original synthetic data, so the attribution of the gain specifically to 'background-agnostic' refinement is under-supported; however, these are experimental validity concerns, not a case of a prediction being equivalent to its input by construction. No load-bearing self-citation chain, imported uniqueness theorem, or ansatz-smuggling citation is present. Since no specific reduction (Equation X = Equation Y by construction, or a fitted parameter renamed as a prediction) can be exhibited from the paper's text, the appropriate finding is no significant circularity.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the accuracy of SAM2 masks, the sufficiency of 12 backgrounds for domain generalization, the validity of the pseudo-labeling in BAR, and the integrity of the user-study design. No new physical entities are introduced.

free parameters (6)
  • k (number of pasted instances) = 3 to 5
    Random range chosen by hand for synthetic composition.
  • max instances per category = 2
    Composition constraint to avoid class imbalance in synthetic images.
  • pairwise IoU limit = 0.5
    Constraint to prevent overlapping pasted instances.
  • BAR second-stage trigger = validation mAP50 > 95.0
    Hand-chosen criterion for starting the white-background fine-tuning phase.
  • detection confidence threshold = 0.4
    Used in the deployed assistant and user study; not in the main mAP evaluation.
  • IoU merge threshold = 0.5
    For multi-frame fusion in the assistant.
assumptions (5)
  • domain assumption SAM2 provides accurate instance masks for the captured parts
    The synthetic dataset is built entirely on SAM2 masks; if masks are imprecise, the composite images contain distorted or incomplete object regions.
  • domain assumption Copy-paste synthetic images with 12 public backgrounds are sufficient to train a detector that generalizes to real factory scenes
    The training set uses only 12 backgrounds not captured from the factory, a domain gap the paper acknowledges in Section III-A.
  • domain assumption The corruption transformations simulate realistic deployment degradations
    Ten hand-selected perturbations with fixed severities are treated as representative of real-world conditions (Section III-D).
  • domain assumption The first-stage model's high-confidence predictions used as BAR pseudo-labels are correct
    The BAR second stage uses high-confidence predictions as labels without verification, as described in Section IV-A.
  • domain assumption The user study's unassisted-first, assisted-second design is not confounded by practice effects
    The unassisted round always precedes the assisted round for each participant, so learning may improve the second round independently of the system (Section V-E).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Snap, Segment, Deploy: A Visual Data and Detection Pipeline for Wearable Industrial Assistants." pith.science (2026). https://pith.science/paper/TZMVYVW5

@misc{pith2026250721072,
  author       = {Pith},
  title        = {Pith review of: Snap, Segment, Deploy: A Visual Data and Detection Pipeline for Wearable Industrial Assistants},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TZMVYVW5}},
  note         = {Machine review of arXiv:2507.21072}
}
read the original abstract

Industrial assembly tasks increasingly demand rapid adaptation to complex procedures and varied components, yet are often conducted in environments with limited computing, connectivity, and strict privacy requirements. These constraints make conventional cloud-based or fully autonomous solutions impractical for factory deployment. This paper introduces a mobile-device-based assistant system for industrial training and operational support, enabling real-time, semi-hands-free interaction through on-device perception and voice interfaces. The system integrates lightweight object detection, speech recognition, and Retrieval-Augmented Generation (RAG) into a modular on-device pipeline that operates entirely on-device, enabling intuitive support for part handling and procedure understanding without relying on manual supervision or cloud services. To enable scalable training, we adopt an automated data construction pipeline and introduce a two-stage refinement strategy to improve visual robustness under domain shift. Experiments on our generated dataset, i.e., Gear8, demonstrate improved robustness to domain shift and common visual corruptions. A structured user study further confirms its practical viability, with positive user feedback on the clarity of the guidance and the quality of the interaction. These results indicate that our framework offers a deployable solution for real-time, privacy-preserving smart assistance in industrial environments. We will release the Gear8 dataset and source code upon acceptance.

Figures

Figures reproduced from arXiv: 2507.21072 by the authors.

Figure 1
Figure 1. Overview of our three-stage pipeline: Snap, Segment, and Deploy. In the Snap stage, we collect multi-angle part videos and real-world background images from industrial environments. In the Segment stage, we extract instance masks using SAM2 [4] and generate a synthetic dataset (Gear8) by compositing parts into factory scenes. In the Deploy stage, a lightweight object detector trained on Gear8 is integrated into a we… view at source ↗
Figure 2
Figure 2. System pipeline of the proposed industrial assistant integrates [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Qualitative detection results on the Gear8 dataset. The first row shows successful object detection cases on the uncorrupted test set, where [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: User-study results for the Smart Assistant system: (a) average scores on the customized functionality-assessment questionnaire (0–10, higher is [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 46 canonical work pages

  1. [1]

    Matteo Capponi, Riccardo Gervasi, Luca Mastrogiacomo, and Fiorenzo Franceschini. Assembly complexity and physiological re- sponse in human-robot collaboration: Insights from a preliminary experimental analysis.Robotics and Computer-Integrated Manufac- turing, 2024

    work page 2024

  2. [2]

    Luca Gualtieri, Maximilian ¨Ohler, Andrea Revolti, and Patrick Dal- lasega. A visual management and augmented-reality-based training module for the enhancement of short and long-term procedural knowl- edge retention in complex machinery setup.Computers & Industrial Engineering, 2024

    work page 2024

  3. [3]

    Lea M Daling and Sabine J Schlittmeier. Effects of augmented reality-, virtual reality-, and mixed reality–based training on objective performance measures and subjective evaluations in manual assembly tasks: a scoping review.Human factors, 2024

    work page 2024

  4. [4]

    Sam 2: Segment anything in images and videos.arXiv preprint arXiv:2408.00714, 2024

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chai- tanya Ryali, Tengyu Ma, Haitham Khedr, Roman R ¨adle, Chloe Rol- land, Laura Gustafson, et al. Sam 2: Segment anything in images and videos.arXiv preprint arXiv:2408.00714, 2024

    arXiv 2024

  5. [5]

    Jinjie Liu, Guolong Liu, Huan Zhao, Junhua Zhao, Jing Qiu, and Zhao Yang Dong. Real-time industrial carbon emission estimation with deep learning-based device recognition and incomplete smart meter data.Engineering Applications of Artificial Intelligence, 127:107272, 2024

    work page 2024

  6. [6]

    Vahid Gharibvand, Mohammad Karimzadeh Kolamroudi, Qasim Zee- shan, Zeki Murat C ¸ ınar, Saeid Sahmani, Mohammed Asmael, and Babak Safaei. Cloud based manufacturing: A review of recent developments in architectures, technologies, infrastructures, platforms and associated challenges.The International Journal of Advanced Manufacturing Technology, 131(1):93–123, 2024

    work page 2024

  7. [7]

    Iar-net: A human-object context guided action recognition network for industrial environment monitoring.IEEE Transactions on Instrumentation and Measurement, 2024

    Naval Kishore Mehta, Shyam Sunder Prasad, Sumeet Saurav, Ravi Saini, and Sanjay Singh. Iar-net: A human-object context guided action recognition network for industrial environment monitoring.IEEE Transactions on Instrumentation and Measurement, 2024

    work page 2024

  8. [8]

    Retrieval-augmented generation for knowledge-intensive nlp tasks.Advances in neural information processing systems, 33:9459–9474, 2020

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich K ¨uttler, Mike Lewis, Wen-tau Yih, Tim Rockt¨aschel, et al. Retrieval-augmented generation for knowledge-intensive nlp tasks.Advances in neural information processing systems, 33:9459–9474, 2020

    work page 2020

Show all 57 references
  1. [9]

    Learning deep object detectors from 3d models

    Xingchao Peng, Baochen Sun, Karim Ali, and Kate Saenko. Learning deep object detectors from 3d models. InProceedings of the IEEE international conference on computer vision, pages 1278–1286, 2015

    2015

  2. [10]

    Playing for data: Ground truth from computer games

    Stephan R Richter, Vibhav Vineet, Stefan Roth, and Vladlen Koltun. Playing for data: Ground truth from computer games. InComputer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part II 14, pages 102–118. Springer, 2016

    2016

  3. [11]

    Playing for benchmarks

    Stephan R Richter, Zeeshan Hayder, and Vladlen Koltun. Playing for benchmarks. InProceedings of the IEEE international conference on computer vision, pages 2213–2222, 2017

    2017

  4. [12]

    Virtual worlds as proxy for multi-object tracking analysis

    Adrien Gaidon, Qiao Wang, Yohann Cabon, and Eleonora Vig. Virtual worlds as proxy for multi-object tracking analysis. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4340–4349, 2016

    2016

  5. [13]

    Unlimited road-scene synthetic annotation (ursa) dataset

    Matt Angus, Mohamed ElBalkini, Samin Khan, Ali Harakeh, Oles An- drienko, Cody Reading, Steven Waslander, and Krzysztof Czarnecki. Unlimited road-scene synthetic annotation (ursa) dataset. In2018 21st International Conference on Intelligent Transportation Systems (ITSC), pages...

    2018

  6. [14]

    Procsy: Procedural synthetic dataset generation towards influence factor stud- ies of semantic segmentation networks

    Samin Khan, Buu Phan, Rick Salay, and Krzysztof Czarnecki. Procsy: Procedural synthetic dataset generation towards influence factor stud- ies of semantic segmentation networks. InCVPR workshops, volume 3, page 4, 2019

    2019

  7. [15]

    Domain randomization for transferring deep neural networks from simulation to the real world

    Josh Tobin, Rachel Fong, Alex Ray, Jonas Schneider, Wojciech Zaremba, and Pieter Abbeel. Domain randomization for transferring deep neural networks from simulation to the real world. In2017 IEEE/RSJ international conference on intelligent robots and systems (IROS), pages 23–30...

    2017

  8. [16]

    Training deep networks with synthetic data: Bridging the reality gap by domain randomization

    Jonathan Tremblay, Aayush Prakash, David Acuna, Mark Brophy, Varun Jampani, Cem Anil, Thang To, Eric Cameracci, Shaad Boo- choon, and Stan Birchfield. Training deep networks with synthetic data: Bridging the reality gap by domain randomization. InPro- ceedings of the IEEE conf...

    2018

  9. [17]

    Object detection using domain randomization and generative adversar- ial refinement of synthetic images.arXiv preprint arXiv:1805.11778, 2018

    Fernando Camaro Nogues, Andrew Huie, and Sakyasingha Dasgupta. Object detection using domain randomization and generative adversar- ial refinement of synthetic images.arXiv preprint arXiv:1805.11778, 2018

    2018 arXiv

  10. [18]

    Synthetic data generation based on rdb-cyclegan for industrial object detection.Mathematics, 11(22):4588, 2023

    Jiwei Hu, Feng Xiao, Qiwen Jin, Guangpeng Zhao, and Ping Lou. Synthetic data generation based on rdb-cyclegan for industrial object detection.Mathematics, 11(22):4588, 2023

    2023

  11. [19]

    Y Lin, K Suzuki, H Takeda, and K Nakamura. Generating synthetic training data for object detection using multi-task generative adversar- ial networks.ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2:443–449, 2020

    2020

  12. [20]

    Bigdatasetgan: Synthesizing imagenet with pixel-wise annotations

    Daiqing Li, Huan Ling, Seung Wook Kim, Karsten Kreis, Sanja Fidler, and Antonio Torralba. Bigdatasetgan: Synthesizing imagenet with pixel-wise annotations. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 21330–21340, 2022

    2022

  13. [21]

    Instagen: Enhancing object detection by training on synthetic dataset

    Chengjian Feng, Yujie Zhong, Zequn Jie, Weidi Xie, and Lin Ma. Instagen: Enhancing object detection by training on synthetic dataset. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 14121–14130, 2024

    2024

  14. [22]

    Cut, paste and learn: Surprisingly easy synthesis for instance detection

    Debidatta Dwibedi, Ishan Misra, and Martial Hebert. Cut, paste and learn: Surprisingly easy synthesis for instance detection. In Proceedings of the IEEE international conference on computer vision, pages 1301–1310, 2017

    2017

  15. [23]

    The synthia dataset: A large collection of synthetic images for semantic segmentation of urban scenes

    German Ros, Laura Sellart, Joanna Materzynska, David Vazquez, and Antonio M Lopez. The synthia dataset: A large collection of synthetic images for semantic segmentation of urban scenes. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 3234...

    2016

  16. [24]

    Rareplanes: Synthetic data takes flight

    Jacob Shermeyer, Thomas Hossler, Adam Van Etten, Daniel Hogan, Ryan Lewis, and Daeil Kim. Rareplanes: Synthetic data takes flight. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 207–217, 2021

    2021

  17. [25]

    Domain adaptive faster r-cnn for object detection in the wild

    Yuhua Chen, Wen Li, Christos Sakaridis, Dengxin Dai, and Luc Van Gool. Domain adaptive faster r-cnn for object detection in the wild. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 3339–3348, 2018

    2018

  18. [26]

    @bench: Benchmarking vision-language models for human-centered assistive technology

    Xin Jiang, Junwei Zheng, Ruiping Liu, Jiahang Li, Jiaming Zhang, Sven Matthiesen, and Rainer Stiefelhagen. @bench: Benchmarking vision-language models for human-centered assistive technology. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (...

    2025

  19. [27]

    Open scene un- derstanding: Grounded situation recognition meets segment anything for helping people with visual impairments

    Ruiping Liu, Jiaming Zhang, Kunyu Peng, Junwei Zheng, Ke Cao, Yufan Chen, Kailun Yang, and Rainer Stiefelhagen. Open scene un- derstanding: Grounded situation recognition meets segment anything for helping people with visual impairments. InProceedings of the IEEE/CVF Internati...

    2023

  20. [28]

    Materobot: Material recognition in wearable robotics for people with visual impairments

    Junwei Zheng, Jiaming Zhang, Kailun Yang, Kunyu Peng, and Rainer Stiefelhagen. Materobot: Material recognition in wearable robotics for people with visual impairments. In2024 IEEE International Conference on Robotics and Automation (ICRA), pages 2303–2309. IEEE, 2024

    2024

  21. [29]

    Ob- jectfinder: Open-vocabulary assistive system for interactive object search by blind people.arXiv preprint arXiv:2412.03118, 2024

    Ruiping Liu, Jiaming Zhang, Angela Sch ¨on, Karin M ¨uller, Junwei Zheng, Kailun Yang, Kathrin Gerling, and Rainer Stiefelhagen. Ob- jectfinder: Open-vocabulary assistive system for interactive object search by blind people.arXiv preprint arXiv:2412.03118, 2024

    2024 arXiv

  22. [30]

    Enhanced yolo-and wearable- based inspection system for automotive wire harness assembly.Ap- plied Sciences, 14(7):2942, 2024

    Shuo Li, Mingzhe Yuan, Wenhong Wang, Feidao Cao, Hongyan Shi, Yuhang Zhang, and Xiangpu Meng. Enhanced yolo-and wearable- based inspection system for automotive wire harness assembly.Ap- plied Sciences, 14(7):2942, 2024

    2024

  23. [31]

    Redundant object detection method for civil aircraft assembly based on machine vision and smart glasses.Measurement Science and Technology, 33(10):105011, 2022

    Qijie Zhao, Yaohui Kong, Shaojie Sheng, and Junjun Zhu. Redundant object detection method for civil aircraft assembly based on machine vision and smart glasses.Measurement Science and Technology, 33(10):105011, 2022

    2022

  24. [32]

    Validating the use of smart glasses in industrial quality control: a case study.Applied Sciences, 14(5):1850, 2024

    Jos ´e Silva, Pedro Coelho, Luzia Saraiva, Paulo Vaz, Pedro Martins, and Alfonso L ´opez-Rivero. Validating the use of smart glasses in industrial quality control: a case study.Applied Sciences, 14(5):1850, 2024

    2024

  25. [33]

    Augmented reality for enhanced visual inspection through knowledge-based deep learning.Structural Health Monitoring, 20(1):426–442, 2021

    Shaohan Wang, Sakib Ashraf Zargar, and Fuh-Gwo Yuan. Augmented reality for enhanced visual inspection through knowledge-based deep learning.Structural Health Monitoring, 20(1):426–442, 2021

    2021

  26. [34]

    An ar-assisted deep learning- based approach for automatic inspection of aviation connectors.IEEE Transactions on Industrial Informatics, 17(3):1721–1731, 2020

    Shufei Li, Pai Zheng, and Lianyu Zheng. An ar-assisted deep learning- based approach for automatic inspection of aviation connectors.IEEE Transactions on Industrial Informatics, 17(3):1721–1731, 2020

    2020

  27. [35]

    Augmented reality maintenance assistant using yolov5.Applied Sciences, 11(11):4758, 2021

    Ana Malta, Mateus Mendes, and Torres Farinha. Augmented reality maintenance assistant using yolov5.Applied Sciences, 11(11):4758, 2021

    2021

  28. [36]

    Deep learning-based smart task assistance in wearable augmented reality.Robotics and Computer-Integrated Manufacturing, 63:101887, 2020

    Kyeong-Beom Park, Minseok Kim, Sung Ho Choi, and Jae Yeol Lee. Deep learning-based smart task assistance in wearable augmented reality.Robotics and Computer-Integrated Manufacturing, 63:101887, 2020

    2020

  29. [37]

    A compatible detector based on improved yolov5 for hydropower device detection in ar inspection system.Expert Systems with Applications, 225:120065, 2023

    Junjie Jiang, Zhile Yang, Chengke Wu, Yuanjun Guo, Meng Yang, and Wei Feng. A compatible detector based on improved yolov5 for hydropower device detection in ar inspection system.Expert Systems with Applications, 225:120065, 2023

    2023

  30. [38]

    Real-time object detection and tracking in mixed reality using microsoft hololens

    Alessandro Farasin, Francesco Peciarolo, Marco Grangetto, Elena Gianaria, Paolo Garza, et al. Real-time object detection and tracking in mixed reality using microsoft hololens. InVISIGRAPP 2020- Proceedings of the 15th International Joint Conference on Computer Vision, Imaging...

    2020

  31. [39]

    A real-time wearable ar system for egocentric vision on the edge

    Iason Karakostas, Aikaterini Valakou, Despoina Gavgiotaki, Zinovia Stefanidi, Ioannis Pastaltzidis, Grigorios Tsipouridis, Nikolaos Kilis, Konstantinos C Apostolakis, Stavroula Ntoa, Nikolaos Dimitriou, et al. A real-time wearable ar system for egocentric vision on the edge. V...

    2024

  32. [40]

    Tracking multiple deformable objects in egocentric videos

    Mingzhen Huang, Xiaoxing Li, Jun Hu, Honghong Peng, and Siwei Lyu. Tracking multiple deformable objects in egocentric videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1461–1471, 2023

    2023

  33. [41]

    Joint hand motion and interaction hotspots prediction from egocentric videos

    Shaowei Liu, Subarna Tripathi, Somdeb Majumdar, and Xiaolong Wang. Joint hand motion and interaction hotspots prediction from egocentric videos. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3282–3292, 2022

    2022

  34. [42]

    A wearable assistive system for the visually impaired using object detection, distance measurement and tactile presentation.Intell

    Yiwen Chen, Junjie Shen, and Hideyuki Sawada. A wearable assistive system for the visually impaired using object detection, distance measurement and tactile presentation.Intell. Robot, 3:420–435, 2023

    2023

  35. [43]

    A google glass based real-time scene analysis for the visually impaired.IEEE Access, 9:166351–166369, 2021

    Sanjeev U Rao, Swaroop Ranganath, TS Ashwin, Guddeti Ram Mo- hana Reddy, et al. A google glass based real-time scene analysis for the visually impaired.IEEE Access, 9:166351–166369, 2021

    2021

  36. [44]

    Underwater smart glasses: A visual-tactile fusion hazard detection system.Iscience, 27(4), 2024

    Zhongze Ma, Chenjie Zhang, and Pengcheng Jiao. Underwater smart glasses: A visual-tactile fusion hazard detection system.Iscience, 27(4), 2024

    2024

  37. [45]

    Egocentric video-language pretraining.Advances in Neural Information Processing Systems, 35:7575–7586, 2022

    Kevin Qinghong Lin, Jinpeng Wang, Mattia Soldan, Michael Wray, Rui Yan, Eric Z Xu, Difei Gao, Rong-Cheng Tu, Wenzhe Zhao, Weijie Kong, et al. Egocentric video-language pretraining.Advances in Neural Information Processing Systems, 35:7575–7586, 2022

    2022

  38. [46]

    Self-adapting large visual-language models to edge devices across visual modalities

    Kaiwen Cai, Zhekai Duan, Gaowen Liu, Charles Fleming, and Chris Xiaoxuan Lu. Self-adapting large visual-language models to edge devices across visual modalities. InEuropean Conference on Computer Vision, pages 301–318. Springer, 2024

    2024

  39. [47]

    An egocentric vision-language model based portable real-time smart assistant.arXiv preprint arXiv:2503.04250, 2025

    Yifei Huang, Jilan Xu, Baoqi Pei, Yuping He, Guo Chen, Mingfang Zhang, Lijin Yang, Zheng Nie, Jinyao Liu, Guoshun Fan, et al. An egocentric vision-language model based portable real-time smart assistant.arXiv preprint arXiv:2503.04250, 2025

    2025 arXiv

  40. [48]

    Yolov11: An overview of the key architectural enhancements.arXiv preprint arXiv:2410.17725, 2024

    Rahima Khanam and Muhammad Hussain. Yolov11: An overview of the key architectural enhancements.arXiv preprint arXiv:2410.17725, 2024

    2024 arXiv

  41. [49]

    Sentence-bert: Sentence embed- dings using siamese bert-networks

    Nils Reimers and Iryna Gurevych. Sentence-bert: Sentence embed- dings using siamese bert-networks. InProceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 11 2019

    2019

  42. [50]

    The faiss library

    Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazar ´e, Maria Lomeli, Lucas Hosseini, and Herv ´e J ´egou. The faiss library. 2024

    2024

  43. [51]

    Depth anything: Unleashing the power of large-scale unlabeled data

    Lihe Yang, Bingyi Kang, Zilong Huang, Xiaogang Xu, Jiashi Feng, and Hengshuang Zhao. Depth anything: Unleashing the power of large-scale unlabeled data. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 10371– 10381, 2024

    2024

  44. [52]

    Phi-3 technical report: A highly capable language model locally on your phone.arXiv preprint arXiv:2404.14219, 2024

    Marah Abdin, Jyoti Aneja, Hany Awadalla, Ahmed Awadallah, Am- mar Ahmad Awan, Nguyen Bach, Amit Bahree, Arash Bakhtiari, Jianmin Bao, Harkirat Behl, et al. Phi-3 technical report: A highly capable language model locally on your phone.arXiv preprint arXiv:2404.14219, 2024

    2024 arXiv

  45. [53]

    Robust speech recognition via large- scale weak supervision

    Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. Robust speech recognition via large- scale weak supervision. InInternational conference on machine learning, pages 28492–28518. PMLR, 2023

    2023

  46. [54]

    pyttsx3.https://github.com/nateshmbhat/ pyttsx3, 2024

    nateshmbhat. pyttsx3.https://github.com/nateshmbhat/ pyttsx3, 2024. Python text-to-speech library

    2024

  47. [55]

    Slic- ing aided hyper inference and fine-tuning for small object detection

    Fatih Cagatay Akyon, Sinan Onur Altinuc, and Alptekin Temizel. Slic- ing aided hyper inference and fine-tuning for small object detection. In2022 IEEE international conference on image processing (ICIP), pages 966–970. IEEE, 2022

    2022

  48. [56]

    Development of nasa-tlx (task load index): Results of empirical and theoretical research

    Sandra G Hart and Lowell E Staveland. Development of nasa-tlx (task load index): Results of empirical and theoretical research. InAdvances in psychology, volume 52, pages 139–183. Elsevier, 1988

    1988

  49. [57]

    How high is high? a meta-analysis of nasa-tlx global workload scores

    Rebecca A Grier. How high is high? a meta-analysis of nasa-tlx global workload scores. InProceedings of the human factors and ergonomics society annual meeting, volume 59, pages 1727–1731. Sage Publications Sage CA: Los Angeles, CA, 2015

    2015

Pith tools

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