{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,25]],"date-time":"2026-05-25T12:03:28Z","timestamp":1779710608734,"version":"3.53.1"},"reference-count":35,"publisher":"Springer Science and Business Media LLC","issue":"3","license":[{"start":{"date-parts":[[2026,4,30]],"date-time":"2026-04-30T00:00:00Z","timestamp":1777507200000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,4,30]],"date-time":"2026-04-30T00:00:00Z","timestamp":1777507200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"Swinburne University of Technology"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["J Ambient Intell Human Comput"],"published-print":{"date-parts":[[2026,5]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>\n                    Industrial hazard detection faces significant challenges due to evolving safety regulations, diverse operational scenarios, and limited training data availability. Current approaches rely on predefined hazard categories and static safety rules, requiring extensive retraining when regulations change. To address these limitations, we propose a neural-symbolic framework that decouples hazard detection into two components: (1) a symbolic program representation of safety rules, and (2) a compositional visual reasoning engine based on Meta Module Networks. This separation enables dynamic adaptation to new safety standards without model retraining, while maintaining high detection accuracy. To evaluate our approach, we introduce HazardComp, a new dataset containing 2,006 real-world images annotated with scene graphs, safety rules, and corresponding hazard queries. The dataset spans diverse industrial environments and enables the evaluation across different hazard types. Experimental results demonstrate that our framework achieves higher performance compared to existing methods, with the inference time of 0.05 s and average accuracies of 86.9%, 90.4%, and 91.6% for object-based, relation-based, and logic-based hazard detection respectively. Our framework\u2019s key innovation lies in its ability to reason about previously unseen hazard scenarios through compositional understanding, offering a more flexible and maintainable solution for real-world safety monitoring applications. Code and dataset are available at:\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" xlink:href=\"https:\/\/github.com\/hailingprojects\/CVR-hazardDet\/tree\/patch-1\" ext-link-type=\"uri\">https:\/\/github.com\/hailingprojects\/CVR-hazardDet\/tree\/patch-1<\/jats:ext-link>\n                    .\n                  <\/jats:p>","DOI":"10.1007\/s12652-026-05088-1","type":"journal-article","created":{"date-parts":[[2026,4,30]],"date-time":"2026-04-30T07:50:28Z","timestamp":1777535428000},"page":"979-994","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Neural-symbolic framework for dynamic hazard detection through compositional visual reasoning"],"prefix":"10.1007","volume":"17","author":[{"given":"Shunlin","family":"Lu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hailing","family":"Zhou","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David","family":"Nguyen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mats","family":"Isaksson","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tao","family":"Peng","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"C. P.","family":"Lim","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Saeid","family":"Nahavandi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lei","family":"Wei","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,4,30]]},"reference":[{"key":"5088_CR1","unstructured":"Hudson D, Manning C (2019) Learning by abstraction: The neural state machine. In: Adv Neural Inf Process Syst 32:"},{"key":"5088_CR2","unstructured":"Sankarasubramanian P, Ganesh E (2021) Artificial intelligence-based detection system for hazardous liquid metal fire. In: 2021 8th International Conference on Computing for Sustainable Global Development (INDIACom), pp. 1\u20136"},{"key":"5088_CR3","doi-asserted-by":"crossref","unstructured":"Avazov K, Mukhiddinov M, Makhmudov F, Cho Y (2022) Fire detection method in smart city environments using a deep-learning-based approach. Electronics 11(1)","DOI":"10.3390\/electronics11010073"},{"issue":"1","key":"5088_CR4","doi-asserted-by":"publisher","first-page":"1671","DOI":"10.32604\/cmc.2023.035762","volume":"75","author":"N Kwak","year":"2023","unstructured":"Kwak N, Kim D (2023) Detection of worker\u2019s safety helmet and mask and identification of worker using deeplearning. Comput Mater Continua 75(1):1671\u20131686","journal-title":"Comput Mater Continua"},{"issue":"6","key":"5088_CR5","doi-asserted-by":"publisher","first-page":"6783","DOI":"10.1007\/s12652-021-03541-x","volume":"14","author":"A Kumar","year":"2023","unstructured":"Kumar A, Kalia A, Sharma A, Kaushal M (2023) A hybrid tiny yolo v4-spp module based improved face mask detection vision system. J Ambient Intell Humaniz Comput 14(6):6783\u20136796. https:\/\/doi.org\/10.1007\/s12652-021-03541-x","journal-title":"J Ambient Intell Humaniz Comput"},{"issue":"12","key":"5088_CR6","doi-asserted-by":"publisher","first-page":"5751","DOI":"10.1007\/s12652-021-03250-5","volume":"13","author":"SJ Berlin","year":"2022","unstructured":"Berlin SJ, John M (2022) Vision based human fall detection with siamese convolutional neural networks. J Ambient Intell Humaniz Comput 13(12):5751\u20135762. https:\/\/doi.org\/10.1007\/s12652-021-03250-5","journal-title":"J Ambient Intell Humaniz Comput"},{"key":"5088_CR7","doi-asserted-by":"publisher","first-page":"104580","DOI":"10.1016\/j.autcon.2022.104580","volume":"144","author":"Y Ding","year":"2022","unstructured":"Ding Y, Liu M, Luo X (2022) Safety compliance checking of construction behaviors using visual question answering. Autom Constr 144:104580. https:\/\/doi.org\/10.1016\/j.autcon.2022.104580","journal-title":"Autom Constr"},{"key":"5088_CR8","doi-asserted-by":"publisher","unstructured":"Li R, Zhang S, He X (2022) Sgtr: End-to-end scene graph generation with transformer. In: 2022 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 19464\u201319474. https:\/\/doi.org\/10.1109\/CVPR52688.2022.01888","DOI":"10.1109\/CVPR52688.2022.01888"},{"key":"5088_CR9","doi-asserted-by":"publisher","unstructured":"Dong L, Lapata M (2018) Coarse-to-fine decoding for neural semantic parsing. In: Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 731\u2013742 . https:\/\/doi.org\/10.18653\/v1\/P18-1068","DOI":"10.18653\/v1\/P18-1068"},{"key":"5088_CR10","doi-asserted-by":"publisher","unstructured":"Chen W, Gan Z, Li L, Cheng Y, Wang W, Liu J (2021) Meta module network for compositional visual reasoning. In: 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 655\u2013664. https:\/\/doi.org\/10.1109\/WACV48630.2021.00070","DOI":"10.1109\/WACV48630.2021.00070"},{"key":"5088_CR11","doi-asserted-by":"publisher","unstructured":"Liu Y, Jiang W (2021) Detection of wearing safety helmet for workers based on yolov4. In: 2021 International Conference on Computer Engineering and Artificial Intelligence (ICCEAI), pp. 83\u201387. https:\/\/doi.org\/10.1109\/ICCEAI52939.2021.00016","DOI":"10.1109\/ICCEAI52939.2021.00016"},{"key":"5088_CR12","doi-asserted-by":"publisher","unstructured":"Zhou F, Zhao H, Nie Z (2021) Safety helmet detection based on yolov5. In: Proceedings of ICPECA, pp. 6\u201311. https:\/\/doi.org\/10.1109\/ICPECA51329.2021.9362711","DOI":"10.1109\/ICPECA51329.2021.9362711"},{"key":"5088_CR13","doi-asserted-by":"publisher","unstructured":"Serna A, Yu X, Saniie J (2024) Ai-based security surveillance and hazard detection for train platform safety. In: Proceedings of eIT, pp. 185\u2013190. https:\/\/doi.org\/10.1109\/eIT60633.2024.10609931","DOI":"10.1109\/eIT60633.2024.10609931"},{"key":"5088_CR14","unstructured":"Murshed M, Verenich E, Gende C, Carroll J, Khan N, Hussain F (2020) Hazard Detection in Supermarkets using Deep Learning on the Edge. arxiv:2003.04116"},{"key":"5088_CR15","doi-asserted-by":"publisher","first-page":"103448","DOI":"10.1016\/j.autcon.2020.103448","volume":"121","author":"I Jeelani","year":"2021","unstructured":"Jeelani I, Asadi K, Ramshankar H, Han K, Albert A (2021) Real-time vision-based worker localization & hazard detection for construction. Autom Constr 121:103448","journal-title":"Autom Constr"},{"key":"5088_CR16","doi-asserted-by":"publisher","first-page":"101152","DOI":"10.1016\/j.aei.2020.101152","volume":"46","author":"B Zhong","year":"2020","unstructured":"Zhong B, Pan X, Love P, Sun J, Tao C (2020) Hazard analysis: A deep learning and text mining framework for accident prevention. Adv Eng Inform 46:101152. https:\/\/doi.org\/10.1016\/j.aei.2020.101152","journal-title":"Adv Eng Inform"},{"issue":"5","key":"5088_CR17","doi-asserted-by":"publisher","first-page":"4697","DOI":"10.1109\/TITS.2023.3240104","volume":"24","author":"D Xiao","year":"2023","unstructured":"Xiao D, Dianati M, Geiger W, Woodman R (2023) Review of graph-based hazardous event detection methods for autonomous driving systems. IEEE Trans Intell Transp Syst 24(5):4697\u20134715. https:\/\/doi.org\/10.1109\/TITS.2023.3240104","journal-title":"IEEE Trans Intell Transp Syst"},{"key":"5088_CR18","doi-asserted-by":"publisher","unstructured":"Abrecht S, Hirsch A, Raafatnia S, Woehrle M (2024) Deep learning safety concerns in automated driving perception. IEEE Transactions on Intelligent Vehicles 1\u201312. https:\/\/doi.org\/10.1109\/TIV.2024.3428415","DOI":"10.1109\/TIV.2024.3428415"},{"key":"5088_CR19","doi-asserted-by":"publisher","first-page":"103310","DOI":"10.1016\/j.autcon.2020.103310","volume":"119","author":"W Fang","year":"2020","unstructured":"Fang W, Ma L, Love HP, Luo D, Zhou L (2020) Knowledge graph for identifying hazards on construction sites: integrating computer vision with ontology. Autom Constr 119:103310. https:\/\/doi.org\/10.1016\/j.autcon.2020.103310","journal-title":"Autom Constr"},{"key":"5088_CR20","doi-asserted-by":"crossref","unstructured":"Kong F, Ahn S (2024) Use of knowledge graphs for construction safety management: a systematic literature review. Information 15(7)","DOI":"10.3390\/info15070390"},{"key":"5088_CR21","doi-asserted-by":"publisher","unstructured":"Wu W, Yuan Q, Chen Q, Cao Y (2023) Construction safety knowledge graph integrating text and image information. In: Proceedings of the 2023 6th International Conference on Information Management and Management Science, pp. 26\u201332. https:\/\/doi.org\/10.1145\/3625469.3625470","DOI":"10.1145\/3625469.3625470"},{"issue":"7","key":"5088_CR22","doi-asserted-by":"publisher","first-page":"7941","DOI":"10.1109\/TITS.2021.3074854","volume":"23","author":"S Yu","year":"2021","unstructured":"Yu S, Malawade A, Muthirayan D, Khargonekar P, Faruque M (2021) Scene-graph augmented data-driven risk assessment of autonomous vehicle decisions. IEEE Trans Intell Transp Syst 23(7):7941\u20137951","journal-title":"IEEE Trans Intell Transp Syst"},{"key":"5088_CR23","doi-asserted-by":"crossref","unstructured":"Ji J, R, K, L, F-F, Niebles J (2020) Action genome: Actions as compositions of spatio-temporal scene graphs. In: Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","DOI":"10.1109\/CVPR42600.2020.01025"},{"issue":"12","key":"5088_CR24","doi-asserted-by":"publisher","first-page":"9379","DOI":"10.1109\/JIOT.2022.3141044","volume":"9","author":"A Malawade","year":"2022","unstructured":"Malawade A, Yu S, Hsu B, Muthirayan D, Khargonekar P, Faruque M (2022) Spatio-temporal scene-graph embedding for autonomous vehicle collision prediction. IEEE Internet Things J 9(12):9379\u20139388","journal-title":"IEEE Internet Things J"},{"key":"5088_CR25","doi-asserted-by":"crossref","unstructured":"Deng J, Yang Z, Chen T, Zhou W, Li H (2021) Transvg: End-to-end visual grounding with transformers. In: 2021 IEEE\/CVF International Conference on Computer Vision (ICCV), pp. 1749\u20131759 . 10.1109\/ICCV48922.2021.00179","DOI":"10.1109\/ICCV48922.2021.00179"},{"key":"5088_CR26","doi-asserted-by":"publisher","unstructured":"Hudson D, Manning C (2019) Gqa: A new dataset for real-world visual reasoning and compositional question answering. In: Proceedings of CVPR, pp. 6693\u20136702. https:\/\/doi.org\/10.1109\/CVPR.2019.00686","DOI":"10.1109\/CVPR.2019.00686"},{"key":"5088_CR27","doi-asserted-by":"publisher","unstructured":"Yu Z, Yu J, Cui Y, Tao D, Tian Q (2019) Deep modular co-attention networks for visual question answering. In: 2019 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 6274\u20136283. https:\/\/doi.org\/10.1109\/CVPR.2019.00644","DOI":"10.1109\/CVPR.2019.00644"},{"key":"5088_CR28","doi-asserted-by":"crossref","unstructured":"Chen Y-C, Li L, Yu L, Kholy AE, Ahmed F, Gan Z, Cheng Y, Liu J (2020) Uniter: Universal image-text representation learning. In: European Conference on Computer Vision (ECCV), vol. 12375. 10.1007\/978-3-030-58577-8_7","DOI":"10.1007\/978-3-030-58577-8_7"},{"key":"5088_CR29","unstructured":"Yi K, Wu J, Gan C, Torralba A, Kohli P, Tenenbaum J (2018) Neural-symbolic vqa: Disentangling reasoning from vision and language understanding. In: Proceedings of NeurIPS, pp. 1039\u20131050"},{"key":"5088_CR30","doi-asserted-by":"publisher","unstructured":"Andreas J, Rohrbach M, Darrell T, Klein D (2016) Learning to compose neural networks for question answering. In: Proceedings of NAACL, pp. 1545\u20131554 . https:\/\/doi.org\/10.18653\/v1\/N16-1181","DOI":"10.18653\/v1\/N16-1181"},{"key":"5088_CR31","doi-asserted-by":"publisher","unstructured":"Johnson J, Hariharan B, Van Der\u00a0Maaten L, Hoffman J, Fei-Fei L, Zitnick C, Girshick R (2017) Inferring and executing programs for visual reasoning. In: Proceedings of ICCV, pp. 3008\u20133017. https:\/\/doi.org\/10.1109\/ICCV.2017.325","DOI":"10.1109\/ICCV.2017.325"},{"key":"5088_CR32","doi-asserted-by":"publisher","unstructured":"Kim E-S, Kang WY, On K-W, Heo Y-J, Zhang B-T (2020) Hypergraph attention networks for multimodal learning. In: IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 14569\u201314578. https:\/\/doi.org\/10.1109\/CVPR42600.2020.01459","DOI":"10.1109\/CVPR42600.2020.01459"},{"key":"5088_CR33","doi-asserted-by":"publisher","first-page":"105095","DOI":"10.1016\/j.imavis.2024.105095","volume":"147","author":"K Su","year":"2024","unstructured":"Su K, Tomioka Y, Zhao Q (2024) Yolic: an efficient method for object localization and classification on edge devices. Image Vis Comput 147:105095. https:\/\/doi.org\/10.1016\/j.imavis.2024.105095","journal-title":"Image Vis Comput"},{"key":"5088_CR34","doi-asserted-by":"publisher","unstructured":"Johnson J, Hariharan B, Maaten L, Fei-Fei L, Zitnick C, R, G (2017) Clevr: a diagnostic dataset for compositional language and elementary visual reasoning. In: Proceedings of CVPR. https:\/\/doi.org\/10.1109\/CVPR.2017.215","DOI":"10.1109\/CVPR.2017.215"},{"key":"5088_CR35","unstructured":"Wang, P., Bai, S., Tan, S., Wang, S., Fan, Z., Bai, J., Chen, K., Liu, X., Wang, J., Ge, W., Fan, Y., Dang, K., Du, M., Ren, X., Men, R., Liu, D., Zhou, C., Zhou, J., Lin, J.: Qwen2-vl: Enhancing vision-language model\u2019s perception of the world at any resolution. arxiv:2409.12191 (2024)"}],"container-title":["Journal of Ambient Intelligence and Humanized Computing"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s12652-026-05088-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s12652-026-05088-1","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s12652-026-05088-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,25]],"date-time":"2026-05-25T11:29:08Z","timestamp":1779708548000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s12652-026-05088-1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,30]]},"references-count":35,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2026,5]]}},"alternative-id":["5088"],"URL":"https:\/\/doi.org\/10.1007\/s12652-026-05088-1","relation":{},"ISSN":["1868-5137","1868-5145"],"issn-type":[{"value":"1868-5137","type":"print"},{"value":"1868-5145","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,30]]},"assertion":[{"value":"28 February 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"10 April 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"30 April 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors declare no Conflict of interest.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}}]}}