{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,8,21]],"date-time":"2025-08-21T16:29:55Z","timestamp":1755793795630,"version":"3.44.0"},"reference-count":33,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,8,15]],"date-time":"2025-08-15T00:00:00Z","timestamp":1755216000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Comput. Sci."],"abstract":"<jats:p>The risk of fires in both indoor and outdoor scenarios is constantly rising around the world. The primary goal of a fire detection system is to minimize financial losses and human casualties by rapidly identifying flames in diverse settings, such as buildings, industrial sites, forests, and rural areas. Traditional fire detection systems that use point sensors have limitations in identifying early ignition and fire spread. Numerous existing computer vision and artificial intelligence-based fire detection techniques have produced good detection rates, but at the expense of excessive false alarms. In this paper, we propose an advanced fire and smoke detection system on the DetectNet_v2 architecture with ResNet-18 as its backbone. The framework uses NVIDIA\u2019s Train-Adapt-Optimize (TAO) transfer learning methods to perform model optimization. We began by curating a custom data set comprising 3,000 real-world and synthetically augmented fire and smoke images to enhance models\u2019 generalization across diverse industrial scenarios. To enable deployment on edge devices, the baseline FP32 model is fine-tuned, pruned, and subsequently optimized using Quantization-Aware Training (QAT) to generate an INT8 precision inference model with its size reduced by 12.7%. The proposed system achieved a detection accuracy of 95.6% for fire and 92% for smoke detections, maintaining a mean inference time of 42\u202fms on RTX GPUs. The comparative analysis revealed that our proposed model outperformed the baseline YOLOv8, SSD MobileNet_v2, and Faster R-CNN models in terms of precision and F1-scores. Performance benchmarks on fire instances such as mAP@0.5 (94.9%), mAP@0.5:0.95 (87.4%), and a low false rate of 3.5% highlight the DetectNet_v2 framework\u2019s robustness and superior detection performance. Further validation experiments on NVIDIA Jetson Orin Nano and Xavier NX platforms confirmed their effective real-time inference capabilities, making them suitable for deployment in safety-critical scenarios and enabling human-in-the-loop verification for efficient alert handling.<\/jats:p>","DOI":"10.3389\/fcomp.2025.1636758","type":"journal-article","created":{"date-parts":[[2025,8,15]],"date-time":"2025-08-15T06:13:03Z","timestamp":1755238383000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Real-time fire and smoke detection system for diverse indoor and outdoor industrial environmental conditions using a vision-based transfer learning approach"],"prefix":"10.3389","volume":"7","author":[{"given":"Uttam U.","family":"Deshpande","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Goh Kah Ong","family":"Michael","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Sufola Das Chagas Silva","family":"Araujo","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Sowmyashree H.","family":"Srinivasaiah","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Harshel","family":"Malawade","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yash","family":"Kulkarni","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yash","family":"Desai","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2025,8,15]]},"reference":[{"key":"ref1","year":"2022"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"73","DOI":"10.3390\/electronics11010073","article-title":"Fire detection method in Smart City environments using a deep-learning-based approach","volume":"11","author":"Avazov","year":"2022","journal-title":"Electronics"},{"key":"ref3","volume-title":"Deep Learning Spring 2025: CIFAR 10 classification","author":"Beji","year":"2025"},{"key":"ref4","doi-asserted-by":"publisher","first-page":"1827","DOI":"10.1016\/j.dsp.2013.07.003","article-title":"Video fire detection \u2013 review","volume":"23","author":"\u00c7etin","year":"2013","journal-title":"Digit. Signal Process."},{"key":"ref5","doi-asserted-by":"publisher","first-page":"1582257","DOI":"10.3389\/frai.2025.1582257","article-title":"Computer-vision based automatic rider helmet violation detection and vehicle identification in Indian smart city scenarios using NVIDIA TAO toolkit and YOLOv8","volume":"8","author":"Deshpande","year":"","journal-title":"Front. Artif. Intell."},{"key":"ref6","doi-asserted-by":"publisher","first-page":"1535775","DOI":"10.3389\/fcomp.2025.1535775","article-title":"Computer vision and AI-based cell phone usage detection in restricted zones of manufacturing industries","volume":"7","author":"Deshpande","year":"","journal-title":"Front. Comput. Sci."},{"key":"ref7","doi-asserted-by":"publisher","first-page":"104","DOI":"10.1007\/s44163-025-00263-3","article-title":"Automatic two-wheeler rider identification and triple-riding detection in surveillance systems using deep-learning models","volume":"5","author":"Deshpande","year":"","journal-title":"Discov. Artif. Intell."},{"key":"ref8","doi-asserted-by":"publisher","first-page":"22045","DOI":"10.1088\/1742-6596\/1237\/2\/022045","article-title":"Population statistics algorithm based on MobileNet_v2","volume":"1237","author":"Feng","year":"2016","journal-title":"J. Physics Conf. Series"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"725","DOI":"10.1109\/TII.2017.2757457","article-title":"Fast smoke detection for video surveillance using CUDA","volume":"14","author":"Filonenko","year":"2018","journal-title":"IEEE Trans. Ind. Inform."},{"key":"ref10","year":"2020"},{"key":"ref11","doi-asserted-by":"publisher","first-page":"2098","DOI":"10.3390\/en13082098","article-title":"AdViSED: advanced video smoke detection for real-time measurements in antifire indoor and outdoor systems","volume":"13","author":"Gagliardi","year":"2020","journal-title":"Energies"},{"key":"ref12","year":"2020"},{"key":"ref13","author":"Healey","year":"1993"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"418","DOI":"10.1016\/j.procs.2020.04.044","article-title":"Low-complexity high-performance deep learning model for real-time low-cost embedded fire detection systems","volume":"171","author":"Jadon","year":"2020","journal-title":"Procedia Comput. Sci."},{"key":"ref15","doi-asserted-by":"publisher","first-page":"e0300502","DOI":"10.1371\/journal.pone.0300502","article-title":"Fire and smoke real-time detection algorithm for coal mines based on improved YOLOv8s","volume":"19","author":"Kong","year":"2024","journal-title":"PLoS One"},{"key":"ref16","doi-asserted-by":"publisher","first-page":"84","DOI":"10.1145\/3065386","article-title":"ImageNet classification with deep convolutional neural networks","volume":"6","author":"Krizhevsky","year":"2017","journal-title":"Commun. ACM"},{"key":"ref17","doi-asserted-by":"publisher","first-page":"1419","DOI":"10.1109\/tsmc.2018.2830099","article-title":"Efficient deep CNN-based fire detection and localization in video surveillance applications","volume":"49","author":"Muhammad","year":"2019","journal-title":"IEEE Trans Syst Man Cybern Syst"},{"key":"ref18","year":"2025"},{"key":"ref19","doi-asserted-by":"publisher","first-page":"105154","DOI":"10.1016\/j.jobe.2022.105154","article-title":"Indoor fire detection utilizing computer vision-based strategies","volume":"61","author":"Pincott","year":"2022","journal-title":"J. Building Eng."},{"key":"ref20","author":"Redmon","year":"2016"},{"key":"ref21","author":"Sandler","year":"2018"},{"key":"ref22","doi-asserted-by":"publisher","first-page":"889","DOI":"10.1007\/s11554-020-01044-0","article-title":"Real-time video fire\/smoke detection based on CNN in antifire surveillance systems","volume":"18","author":"Saponara","year":"2021","journal-title":"J. Real-Time Image Proc."},{"key":"ref23","doi-asserted-by":"publisher","first-page":"165","DOI":"10.1186\/s42408-022-00165-0","article-title":"Forest fire and smoke detection using deep learning-based learning without forgetting","volume":"19","author":"Sathishkumar","year":"2023","journal-title":"Fire Ecol."},{"key":"ref24","volume-title":"BoWFire, Kaggle","author":"Senthil","year":"2025"},{"key":"ref25","doi-asserted-by":"publisher","first-page":"1","DOI":"10.13044\/j.sdewes.d8.0378","article-title":"Occupancy heat gain and prediction using deep learning approach for reducing building energy demand","volume":"9","author":"Tien","year":"2020","journal-title":"J. Sustain. Develop. Energy Water Environ. Syst."},{"key":"ref26","doi-asserted-by":"publisher","first-page":"10481","DOI":"10.1007\/s00500-019-04557-4","article-title":"A new artificial bee colony algorithm-based color space for fire\/flame detection","volume":"24","author":"Topta\u015f","year":"2020","journal-title":"Soft. Comput."},{"key":"ref27","volume-title":"Review: G-RMI - winner in 2016 COCO detection (object detection)","author":"Tsang","year":"2020"},{"key":"ref28","doi-asserted-by":"publisher","first-page":"3","DOI":"10.1016\/j.patrec.2012.07.005","article-title":"Intelligent multi-camera video surveillance: A review","volume":"34","author":"Wang","year":"2013","journal-title":"Pattern Recogn. Lett."},{"key":"ref29","doi-asserted-by":"publisher","first-page":"115506","DOI":"10.1016\/j.apenergy.2020.115506","article-title":"Vision-based detection and prediction of equipment heat gains in commercial office buildings using a deep learning method","volume":"277","author":"Wei","year":"2020","journal-title":"Appl. Energy"},{"key":"ref30","author":"Wu","year":"2018"},{"key":"ref31","volume-title":"Video-based smoke detection: Possibilities, techniques, and challenges. Presented at the IFPA, fire suppression and detection research and applications\u2014A technical working conference (SUPDET)","author":"Xiong","year":"2007"},{"key":"ref32","author":"Zhang","year":"2016"},{"key":"ref33","doi-asserted-by":"publisher","first-page":"50","DOI":"10.1016\/j.firesaf.2016.08.004","article-title":"Wildfire smoke detection based on local extremal region segmentation and surveillance","volume":"85","author":"Zhou","year":"2016","journal-title":"Fire Saf. J."}],"container-title":["Frontiers in Computer Science"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2025.1636758\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,8,15]],"date-time":"2025-08-15T06:13:04Z","timestamp":1755238384000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2025.1636758\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,8,15]]},"references-count":33,"alternative-id":["10.3389\/fcomp.2025.1636758"],"URL":"https:\/\/doi.org\/10.3389\/fcomp.2025.1636758","relation":{},"ISSN":["2624-9898"],"issn-type":[{"value":"2624-9898","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,8,15]]},"article-number":"1636758"}}