{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,18]],"date-time":"2026-08-18T14:08:50Z","timestamp":1787062130799,"version":"build-2736575974"},"reference-count":32,"publisher":"MDPI AG","issue":"6","license":[{"start":{"date-parts":[[2026,3,11]],"date-time":"2026-03-11T00:00:00Z","timestamp":1773187200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Economic Revitalization and Resilience Enhancement Plan \u201cNew Generation Lithuania\u201d","award":["02-002-P-0001"],"award-info":[{"award-number":["02-002-P-0001"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>This study investigates efficient thermal-image human pose estimation under edge deployment constraints for smart home activity recognition. A single-person thermal dataset of 2500 images was collected and annotated with 17 body keypoints. YOLO11-pose and YOLOv8-pose models were trained and evaluated across all five model scales (n\u2013x) at three input resolutions 640 \u00d7 512, 320 \u00d7 256, and 160 \u00d7 128 px. The accuracy was evaluated using box mean Average Precision (mAP50\u201395), pose mAP50\u201395, and Object Keypoint Similarity (OKS) metrics. Runtime performance was assessed using per-image latency and power measurements on three NVIDIA Jetson platforms: Orin Nano 4 GB, Orin Nano 8 GB and AGX Orin 64 GB, using PyTorch and TensorRT at FP32, FP16, INT8 precision. Human detection remained consistently high across model variants, whereas pose accuracy decreased as the input resolution was reduced. TensorRT FP16 preserved pose accuracy relative to PyTorch and TensorRT FP32, with minimal changes in OKS and pose mAP50\u201395, while notably reducing per-image latency and power consumption. INT8 further reduced power consumption and in some configurations improved latency, but caused configuration-dependent losses in OKS and pose mAP50\u201395. The findings indicate that FP16 offers the best accuracy\u2013efficiency balance for thermal pose estimation on edge devices, while practical feasibility depends on device capabilities and memory limitations.<\/jats:p>","DOI":"10.3390\/s26061774","type":"journal-article","created":{"date-parts":[[2026,3,11]],"date-time":"2026-03-11T14:43:18Z","timestamp":1773240198000},"page":"1774","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["Efficient Thermal Pose Estimation: Balancing Accuracy and Edge Deployment for Smart Home Activity Recognition"],"prefix":"10.3390","volume":"26","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-5714-9188","authenticated-orcid":false,"given":"Gabriela","family":"Vdoviak","sequence":"first","affiliation":[{"name":"Department of Electronic Systems, Vilnius Gediminas Technical University, Saul\u0117tekio Ave. 11, LT-10223 Vilnius, Lithuania"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3538-218X","authenticated-orcid":false,"given":"Tomyslav","family":"Sledevi\u010d","sequence":"additional","affiliation":[{"name":"Department of Electronic Systems, Vilnius Gediminas Technical University, Saul\u0117tekio Ave. 11, LT-10223 Vilnius, Lithuania"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1588-6572","authenticated-orcid":false,"given":"Vytautas","family":"Abromavi\u010dius","sequence":"additional","affiliation":[{"name":"Department of Electronic Systems, Vilnius Gediminas Technical University, Saul\u0117tekio Ave. 11, LT-10223 Vilnius, Lithuania"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8897-7366","authenticated-orcid":false,"given":"Dalius","family":"Navakauskas","sequence":"additional","affiliation":[{"name":"Department of Electronic Systems, Vilnius Gediminas Technical University, Saul\u0117tekio Ave. 11, LT-10223 Vilnius, Lithuania"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9800-9158","authenticated-orcid":false,"given":"Art\u016bras","family":"Kaklauskas","sequence":"additional","affiliation":[{"name":"Department of Construction Management and Real Estate, Vilnius Gediminas Technical University, Saul\u0117tekio Ave. 11, LT-10223 Vilnius, Lithuania"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,3,11]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"10825","DOI":"10.1007\/s13369-023-07720-0","article-title":"A new home-based upper-and lower-limb telerehabilitation platform with experimental validation","volume":"48","author":"Latreche","year":"2023","journal-title":"Arab. J. Sci. Eng."},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Clemente, C., Chambel, G., Silva, D.C., Montes, A.M., Pinto, J.F., and Silva, H.P.D. (2023). Feasibility of 3D body tracking from monocular 2D video feeds in musculoskeletal telerehabilitation. Sensors, 24.","DOI":"10.3390\/s24010206"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Aguilar-Ortega, R., Berral-Soler, R., Jim\u00e9nez-Velasco, I., Romero-Ram\u00edrez, F.J., Garc\u00eda-Mar\u00edn, M., Zafra-Palma, J., Mu\u00f1oz-Salinas, R., Medina-Carnicer, R., and Mar\u00edn-Jim\u00e9nez, M.J. (2023). Uco physical rehabilitation: New dataset and study of human pose estimation methods on physical rehabilitation exercises. Sensors, 23.","DOI":"10.3390\/s23218862"},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"68","DOI":"10.1007\/s10462-024-11060-2","article-title":"A systematic survey on human pose estimation: Upstream and downstream tasks, approaches, lightweight models, and prospects","volume":"58","author":"Gao","year":"2025","journal-title":"Artif. Intell. Rev."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"85","DOI":"10.1080\/20476965.2024.2395574","article-title":"Next-generation fall detection: Harnessing human pose estimation and transformer technology","volume":"14","author":"Sykes","year":"2025","journal-title":"Health Syst."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"102209","DOI":"10.1016\/j.inffus.2023.102209","article-title":"In-bed human pose estimation using multi-source information fusion for health monitoring in real-world scenarios","volume":"105","author":"Zhu","year":"2024","journal-title":"Inf. Fusion"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"126804","DOI":"10.1016\/j.neucom.2023.126804","article-title":"A systematic literature review on object detection using near infrared and thermal images","volume":"560","author":"Bustos","year":"2023","journal-title":"Neurocomputing"},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"114215","DOI":"10.1016\/j.jobe.2025.114215","article-title":"Vision-based Occupancy Detection in Indoor Environments: A Comparison of Standard RGB and Thermal Cameras","volume":"114","author":"Zhang","year":"2025","journal-title":"J. Build. Eng."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Guo, Y., Chen, Y., Deng, J., Li, S., and Zhou, H. (2022). Identity-preserved human posture detection in infrared thermal images: A benchmark. Sensors, 23.","DOI":"10.3390\/s23010092"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Ficili, I., Giacobbe, M., Tricomi, G., and Puliafito, A. (2025). From sensors to data intelligence: Leveraging IoT, cloud, and edge computing with AI. Sensors, 25.","DOI":"10.3390\/s25061763"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Liu, J., Du, Y., Yang, K., Wu, J., Wang, Y., Hu, X., Wang, Z., Liu, Y., Sun, P., and Boukerche, A. (2025). Edge-cloud collaborative computing on distributed intelligence and model optimization: A survey. arXiv.","DOI":"10.1109\/COMST.2026.3669216"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Cajas Ord\u00f3\u00f1ez, S.A., Samanta, J., Su\u00e1rez-Cetrulo, A.L., and Carbajo, R.S. (2025). Intelligent edge computing and machine learning: A survey of optimization and applications. Future Internet, 17.","DOI":"10.3390\/fi17090417"},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Zeng, L., Huang, C., Xie, R., Huang, Z., Guo, Y., He, L., Xie, Z., and Xing, G. (2025). ThermiKit: Edge-Optimized LWIR Analytics with Agent-Driven Interactions. Proceedings of the 2025 ACM International Workshop on Thermal Sensing and Computing, Association for Computing Machinery.","DOI":"10.1145\/3737905.3769284"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"35352","DOI":"10.1109\/ACCESS.2023.3264714","article-title":"Human pose estimation using thermal images","volume":"11","author":"Smith","year":"2023","journal-title":"IEEE Access"},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"174964","DOI":"10.1109\/ACCESS.2020.3025413","article-title":"Multi-person pose estimation using thermal images","volume":"8","author":"Chen","year":"2020","journal-title":"IEEE Access"},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"1757","DOI":"10.1007\/s11760-021-01916-3","article-title":"Pose estimation at night in infrared images using a lightweight multi-stage attention network","volume":"15","author":"Zang","year":"2021","journal-title":"Signal Image Video Process."},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Cormier, M., Yi, C.N.Z., Specker, A., Bla\u00df, B., Heizmann, M., and Beyerer, J. (2024). Leveraging Thermal Imaging for Robust Human Pose Estimation in Low-Light Vision. Proceedings of the Asian Conference on Computer Vision, Springer Nature.","DOI":"10.1007\/978-981-96-2641-0_5"},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"102154","DOI":"10.1016\/j.inffus.2023.102154","article-title":"3D Human Pose Estimation from multi-view thermal vision sensors","volume":"104","author":"Sanjuan","year":"2024","journal-title":"Inf. Fusion"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Tambwekar, A., Park, B.K.D., Kusari, A., and Sun, W. (2024). Three-Dimensional Posture Estimation of Vehicle Occupants Using Depth and Infrared Images. Sensors, 24.","DOI":"10.3390\/s24175530"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"124089","DOI":"10.1016\/j.eswa.2024.124089","article-title":"Real-time multi-camera 3D human pose estimation at the edge for industrial applications","volume":"252","author":"Boldo","year":"2024","journal-title":"Expert Syst. Appl."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Analia, R., Forster, A., Xie, S.Q., and Zhang, Z. (2025). Privacy-Preserving Approach for Early Detection of Long-Lie Incidents: A Pilot Study with Healthy Subjects. Sensors, 25.","DOI":"10.3390\/s25123836"},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"2849","DOI":"10.1007\/s12273-025-1335-6","article-title":"A novel lightweight skeletal temporal model for real-time, computationally efficient recognition of occupant thermal adaptation behavior","volume":"Volume 18","author":"Wang","year":"2025","journal-title":"Building Simulation"},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"902","DOI":"10.1109\/TBIOM.2025.3575499","article-title":"OpenThermalPose2: Extending the Open-Source Annotated Thermal Human Pose Dataset With More Data, Subjects, and Poses","volume":"7","author":"Kuzdeuov","year":"2025","journal-title":"IEEE Trans. Biom. Behav. Identity Sci."},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"Gad, M.M., Gad, W., Abdelkader, T., and Naik, K. (2025). Personalized Smart Home Automation Using Machine Learning: Predicting User Activities. Sensors, 25.","DOI":"10.3390\/s25196082"},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"60822","DOI":"10.1109\/ACCESS.2025.3556001","article-title":"Human Daily Indoor Action (HDIA) Dataset: Privacy-Preserving Human Action Recognition Using Infrared Camera and Wearable Armband Sensors","volume":"13","author":"Park","year":"2025","journal-title":"IEEE Access"},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Iadarola, G., Mengarelli, A., Iarlori, S., Monteri\u00f9, A., and Spinsante, S. (2025). RGB-D Cameras and Brain\u2013Computer Interfaces for Human Activity Recognition: An Overview. Sensors, 25.","DOI":"10.3390\/s25206286"},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Cormier, M., Specker, A., and Beyerer, J. (2025). UPPET: Unified Pedestrian Pose Estimation in Thermal Imaging. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), IEEE Computer Society.","DOI":"10.1109\/CVPRW67362.2025.00440"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"37175","DOI":"10.1109\/JIOT.2025.3582636","article-title":"RailVoxelDet: An Lightweight 3D Object Detection Method for Railway Transportation Driven by on-Board LiDAR Data","volume":"12","author":"Chen","year":"2025","journal-title":"IEEE Internet Things J."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"113923","DOI":"10.1016\/j.ymssp.2026.113923","article-title":"A liquid-impulse neural network model based on heterogeneous fusion of multimodal information for interpretable rotating machinery fault diagnosis","volume":"246","author":"You","year":"2026","journal-title":"Mech. Syst. Signal Process."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"102435","DOI":"10.1016\/j.media.2022.102435","article-title":"Dual-channel cascade pose estimation network trained on infrared thermal image and groundtruth annotation for real-time gait measurement","volume":"79","author":"Zhu","year":"2022","journal-title":"Med. Image Anal."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"104922","DOI":"10.1016\/j.dsp.2024.104922","article-title":"A low-resolution infrared gesture recognition method combining weak information reconstruction and joint training strategy","volume":"158","author":"Chen","year":"2025","journal-title":"Digit. Signal Process."},{"key":"ref_32","doi-asserted-by":"crossref","unstructured":"Wang, Y., Wu, Y., Ogai, H., Dong, Y., and Tateno, S. (2025, January 19\u201321). Edge-Oriented Person Activity Recognition and Alert System for Ensuring Safety of Independent Elderly at Home. Proceedings of the 2025 8th International Conference on Information Communication and Signal Processing (ICICSP), Xi\u2019an, China.","DOI":"10.1109\/ICICSP66564.2025.11338337"}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/6\/1774\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,13]],"date-time":"2026-03-13T05:53:53Z","timestamp":1773381233000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/6\/1774"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,11]]},"references-count":32,"journal-issue":{"issue":"6","published-online":{"date-parts":[[2026,3]]}},"alternative-id":["s26061774"],"URL":"https:\/\/doi.org\/10.3390\/s26061774","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,11]]}}}