{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,15]],"date-time":"2026-06-15T14:27:05Z","timestamp":1781533625644,"version":"3.54.5"},"reference-count":27,"publisher":"MDPI AG","issue":"5","license":[{"start":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T00:00:00Z","timestamp":1778716800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"crossref","award":["41901279"],"award-info":[{"award-number":["41901279"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"crossref","award":["42505175"],"award-info":[{"award-number":["42505175"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"crossref"}]},{"award":["41901279"],"award-info":[{"award-number":["41901279"]}],"id":[{"id":"https:\/\/ror.org\/01h0zpd94","id-type":"ROR","asserted-by":"publisher"}]},{"award":["42505175"],"award-info":[{"award-number":["42505175"]}],"id":[{"id":"https:\/\/ror.org\/01h0zpd94","id-type":"ROR","asserted-by":"publisher"}]},{"name":"Science Foundation Research Project of Wuhan Institute of Technology","award":["K202239"],"award-info":[{"award-number":["K202239"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Computers"],"abstract":"<jats:p>In recent years, wildfires have occurred with increasing frequency. Pixel-level annotation of high-resolution remote sensing wildfire imagery is costly and labor-intensive. Therefore, there is an urgent need for a weakly supervised wildfire detection method that balances detection accuracy and annotation efficiency. To address the key limitations of existing weakly supervised approaches based on class activation maps (CAMs), including imprecise delineation of fire boundaries, insufficient utilization of cross-modal information, and limited capability in modeling temporal characteristics, this paper proposes a dual-branch multi-scale feature fusion framework for weakly supervised wildfire detection. The proposed framework consists of a multispectral branch and a shortwave infrared (SWIR) temporal branch, which are designed to capture the spatial structural information of fire regions and the temporal variation of thermal anomalies, respectively. Attention-guided feature fusion modules are introduced at each network stage to enable complementary integration of cross-modal information. In addition, a multi-scale CAM-weighted fusion strategy is designed to jointly enhance region localization accuracy and semantic discrimination capability. Experimental evaluations are conducted on a high-resolution wildfire dataset covering 29 regions and consisting of 2206 images. The results demonstrate that the proposed method achieves an IoU of 58.7% and an F1-score of 73.5%, outperforming the state-of-the-art methods by 4.6% and 3.2%, respectively. Ablation and comparative experiments further verify that the dual-branch architecture and feature fusion strategy significantly improve fire localization accuracy and effectively reduce the missed detection rate.<\/jats:p>","DOI":"10.3390\/computers15050311","type":"journal-article","created":{"date-parts":[[2026,5,14]],"date-time":"2026-05-14T13:04:09Z","timestamp":1778763849000},"page":"311","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["A Weakly Supervised Multi-Scale Cross-Modal Information Fusion Method for Wildfire Detection"],"prefix":"10.3390","volume":"15","author":[{"given":"Dawei","family":"Wen","sequence":"first","affiliation":[{"name":"School of Computer Science and Engineering, Wuhan Institute of Technology, Wuhan 430205, China"},{"name":"Hubei Provincial Key Laboratory of Intelligent Robot, Wuhan Institute of Technology, Wuhan 430205, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhoujiang","family":"Peng","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Wuhan Institute of Technology, Wuhan 430205, China"},{"name":"Hubei Provincial Key Laboratory of Intelligent Robot, Wuhan Institute of Technology, Wuhan 430205, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yuan","family":"Tian","sequence":"additional","affiliation":[{"name":"School of Geographical Sciences, Lingnan Normal University, Zhanjiang 524048, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,5,14]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"2047","DOI":"10.1111\/1365-2745.13403","article-title":"Fire as a fundamental ecological process: Research advances and frontiers","volume":"108","author":"McLauchlan","year":"2020","journal-title":"J. Ecol."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"641","DOI":"10.1071\/WF18108","article-title":"High wildfire damage in interface communities in California","volume":"28","author":"Kramer","year":"2019","journal-title":"Int. J. Wildland Fire"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Zhu, D., and Yang, P. (2024). Study on the Evolutionary Characteristics of Post-Fire Forest Recovery Using Unmanned Aerial Vehicle Imagery and Deep Learning: A Case Study of Jinyun Mountain in Chongqing, China. Sustainability, 16.","DOI":"10.3390\/su16229717"},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"6803","DOI":"10.1080\/01431161.2025.2546154","article-title":"Evaluating the capabilities of high-resolution PlanetScope and Sentinel-2 images for mapping burned area and vegetation regrowth in Indonesia\u2019s savannas","volume":"46","author":"Vetrita","year":"2025","journal-title":"Int. J. Remote Sens."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"867","DOI":"10.5194\/essd-16-867-2024","article-title":"GloCAB: Global cropland burned area from mid-2002 to 2020","volume":"16","author":"Hall","year":"2024","journal-title":"Earth Syst. Sci. Data"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"2015","DOI":"10.5194\/essd-10-2015-2018","article-title":"Generation and analysis of a new global burned area product based on MODIS 250 m reflectance bands and thermal anomalies","volume":"10","author":"Chuvieco","year":"2018","journal-title":"Earth Syst. Sci. Data"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"113214","DOI":"10.1016\/j.rse.2022.113214","article-title":"Assessment and characterization of sources of error impacting the accuracy of global burned area products","volume":"280","author":"Franquesa","year":"2022","journal-title":"Remote Sens. Environ."},{"key":"ref_8","unstructured":"NASA (2025, January 01). Fire Information for Resource Management System (FIRMS), Available online: https:\/\/firms.modaps.eosdis.nasa.gov\/."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Xiang, J., Xing, Y., Wei, W., Yan, E., Jiang, J., and Mo, D. (2023). Dynamic Detection of Forest Change in Hunan Province Based on Sentinel-2 Images and Deep Learning. Remote Sens., 15.","DOI":"10.3390\/rs15030628"},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"31076","DOI":"10.1038\/s41598-024-81976-w","article-title":"Enhancing active fire detection in Sentinel 2 imagery using GLCM texture features in random forest models","volume":"14","author":"Zhou","year":"2024","journal-title":"Sci. Rep."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"2196370","DOI":"10.1080\/19475705.2023.2196370","article-title":"Uni-temporal Sentinel-2 imagery for wildfire detection using deep learning semantic segmentation models","volume":"14","author":"Ilyas","year":"2023","journal-title":"Geomat. Nat. Hazards Risk"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Hong, Z., Tang, Z., Pan, H., Zhang, Y., Zheng, Z., Zhou, R., Ma, Z., Zhang, Y., Han, Y., and Wang, J. (2022). Active fire detection using a novel convolutional neural network based on Himawari-8 satellite images. Front. Environ. Sci., 10.","DOI":"10.3389\/fenvs.2022.794028"},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Wang, J., Wang, Y., Liu, L., Yin, H., Ye, N., and Xu, C. (2023). Weakly Supervised Forest Fire Segmentation in UAV Imagery Based on Foreground-Aware Pooling and Context-Aware Loss. Remote Sens., 15.","DOI":"10.3390\/rs15143606"},{"key":"ref_14","first-page":"5620211","article-title":"Contrastive Tokens and Label Activation for Remote Sensing Weakly Supervised Semantic Segmentation","volume":"62","author":"Hu","year":"2024","journal-title":"IEEE Trans. Geosci. Remote Sens."},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Chen, Z., Lian, Y., Bai, J., Zhang, J., Xiao, Z., and Hou, B. (2025). Weakly Supervised Semantic Segmentation of Remote Sensing Images Using Siamese Affinity Network. Remote Sens., 17.","DOI":"10.3390\/rs17050808"},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Zhou, B., Khosla, A., Lapedriza, A., Oliva, A., and Torralba, A. (2016). Learning deep features for discriminative localization. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, IEEE.","DOI":"10.1109\/CVPR.2016.319"},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"110556","DOI":"10.1016\/j.engappai.2025.110556","article-title":"Attention Guided Class Activation Maps for Boosting Weakly Supervised Semantic Segmentation","volume":"151","author":"Li","year":"2025","journal-title":"Eng. Appl. Artif. Intell."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Amaral, B., Niknejad, M., Barata, C., and Bernardino, A. (2022). Weakly Supervised Fire and Smoke Segmentation in Forest Images with CAM and CRF. 2022 26th International Conference on Pattern Recognition (ICPR), IEEE.","DOI":"10.1109\/ICPR56361.2022.9956288"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"White, L.A., and Gibson, R.K. (2022). Comparing fire extent and severity mapping between Sentinel-2 and Landsat 8 satellite sensors. Remote Sens., 14.","DOI":"10.3390\/rs14071661"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"4100314","DOI":"10.1109\/TGRS.2025.3647717","article-title":"A General and Robust Active Fire Detection Method Using Sentinel-2 and Landsat Shortwave Infrared Measurements","volume":"64","author":"Gui","year":"2026","journal-title":"IEEE Trans. Geosci. Remote Sens."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Wickramasinghe, C.H., Jones, S., Reinke, K., and Wallace, L. (2016). Development of a multi-spatial resolution approach to the surveillance of active fire lines using Himawari-8. Remote Sens., 8.","DOI":"10.3390\/rs8110932"},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"111254","DOI":"10.1016\/j.rse.2019.111254","article-title":"Landsat-8 and Sentinel-2 burned area mapping: A combined sensor multi-temporal change detection approach","volume":"231","author":"Roy","year":"2019","journal-title":"Remote Sens. Environ."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"5606121","DOI":"10.1109\/TGRS.2025.3528641","article-title":"Wildfire Detection and Mapping by Satellite With an Enhanced Configuration of the Normalized Hotspot Indices: Results From Sentinel-2 and Landsat 8\/9 Data Integration","volume":"63","author":"Mazzeo","year":"2025","journal-title":"IEEE Trans. Geosci. Remote Sens."},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"1854","DOI":"10.11834\/jrs.20242314","article-title":"Progress and prospect of forest fire monitoring based on multi-source remote sensing data","volume":"28","author":"Cao","year":"2024","journal-title":"Natl. Remote Sens. Bull."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"27302","DOI":"10.1109\/JSTARS.2025.3623394","article-title":"Cross-Modal Feature Calibration and Fusion Network for Remote Sensing Optical\u2013SAR Joint Object Detection Under Cloud Occlusion","volume":"18","author":"Ding","year":"2025","journal-title":"IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens."},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"488","DOI":"10.1007\/s12145-025-01991-x","article-title":"Smoke characterization for incipient wildfire detection from Sentinel-2 sensor based on sigmoid activation function and momentum gradient descent optimizer","volume":"18","author":"Sali","year":"2025","journal-title":"Earth Sci. Inform."},{"key":"ref_27","unstructured":"Loschilov, I., and Hutter, F. (2019). Decoupled weight decay regularization. arXiv."}],"container-title":["Computers"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2073-431X\/15\/5\/311\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,16]],"date-time":"2026-05-16T04:19:31Z","timestamp":1778905171000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2073-431X\/15\/5\/311"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,14]]},"references-count":27,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2026,5]]}},"alternative-id":["computers15050311"],"URL":"https:\/\/doi.org\/10.3390\/computers15050311","relation":{},"ISSN":["2073-431X"],"issn-type":[{"value":"2073-431X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,14]]}}}