{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,25]],"date-time":"2026-08-25T02:58:11Z","timestamp":1787626691421,"version":"build-2736575974"},"reference-count":44,"publisher":"MDPI AG","issue":"4","license":[{"start":{"date-parts":[[2026,2,21]],"date-time":"2026-02-21T00:00:00Z","timestamp":1771632000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100000181","name":"Air Force Office of Scientific Research","doi-asserted-by":"publisher","award":["FA9550-24-1-0237"],"award-info":[{"award-number":["FA9550-24-1-0237"]}],"id":[{"id":"10.13039\/100000181","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100006192","name":"Office of Advanced Scientific Computing Research within the Department of Energy Office of Science","doi-asserted-by":"publisher","award":["DE-SC0023163"],"award-info":[{"award-number":["DE-SC0023163"]}],"id":[{"id":"10.13039\/100006192","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Camera-based wildlife monitoring is often overwhelmed by non-target triggers and slowed by manual review or cloud-dependent inference, which can prevent timely intervention for high stakes human\u2013wildlife conflicts. Our key contribution is a deployable, fully offline edge vision sensor that achieves near-real-time, highly accurate wildlife event classification by combining detector-based empty-image suppression with a lightweight classifier trained with a staged transfer-learning curriculum. Specifically, Stage 1 uses a pretrained You Only Look Once (YOLO)-family detector for permissive animal localization and empty-trigger suppression, and Stage 2 uses a lightweight EfficientNet-based binary classifier to confirm puma on detector crops and gate downstream actions. Our design is robust to low-quality nighttime monochrome imagery (motion blur, low contrast, illumination artifacts, and partial-body captures) and operates using commercially available components in connectivity-limited settings. In field deployments running since May 2025, end-to-end latency from camera trigger to action command is approximately 4 s. Ablation studies using a dataset of labeled wildlife images (pumas, not pumas) show that the two-stage approach substantially reduces false alarms in identifying pumas relative to a full-image classifier while maintaining high recall. On the held-out test set (N=1434 events), the proposed two-stage cascade achieves precision 0.983, recall 0.975, F1 0.979, accuracy 0.986, and balanced accuracy 0.983, with only 8 false positives and 12 false negatives. The system can be easily adapted for other species, as demonstrated by rapid retraining of the second stage to classify ringtails. Downstream responses (e.g., notifications and optional audio\/light outputs) provide flexible actuation capabilities that can be configured to support intervention.<\/jats:p>","DOI":"10.3390\/s26041366","type":"journal-article","created":{"date-parts":[[2026,2,23]],"date-time":"2026-02-23T08:58:21Z","timestamp":1771837101000},"page":"1366","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["Two-Stage Wildlife Event Classification for Edge Deployment"],"prefix":"10.3390","volume":"26","author":[{"ORCID":"https:\/\/orcid.org\/0009-0008-6773-4451","authenticated-orcid":false,"given":"Aditya S.","family":"Viswanathan","sequence":"first","affiliation":[{"name":"Department of Energy Science and Engineering, Stanford University, Stanford, CA 94305, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Adis","family":"Bock","sequence":"additional","affiliation":[{"name":"Pajarito Environmental Education Center, Los Alamos, NM 87544, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zoe","family":"Bent","sequence":"additional","affiliation":[{"name":"Rubenstein School for Environment and Natural Resources, University of Vermont, Burlington, VT 05405, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0006-6607-6544","authenticated-orcid":false,"given":"Mark A.","family":"Peyton","sequence":"additional","affiliation":[{"name":"Department of Fish, Wildlife and Conservation Ecology, New Mexico State University, Las Cruces, NM 88003, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9019-8935","authenticated-orcid":false,"given":"Daniel M.","family":"Tartakovsky","sequence":"additional","affiliation":[{"name":"Department of Energy Science and Engineering, Stanford University, Stanford, CA 94305, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Javier E.","family":"Santos","sequence":"additional","affiliation":[{"name":"Los Alamos National Laboratory, Earth and Environmental Sciences Division, Los Alamos, NM 87545, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,2,21]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"3503","DOI":"10.1002\/ece3.6147","article-title":"Three critical factors affecting automated image species recognition performance for camera traps","volume":"10","author":"Schneider","year":"2020","journal-title":"Ecol. Evol."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"4590","DOI":"10.1038\/s41598-019-40926-7","article-title":"Improving estimation of puma (Puma concolor) population density: Clustered camera-trapping, telemetry data, and generalized spatial mark-resight models","volume":"9","author":"Murphy","year":"2019","journal-title":"Sci. Rep."},{"key":"ref_3","unstructured":"Beery, S. (2023). The MegaDetector: Large-Scale Deployment of Computer Vision for Conservation and Biodiversity Monitoring, California Institute of Technology."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1017\/S0376892919000298","article-title":"Wildlife insights: A platform to maximize the potential of camera trap and other passive sensor wildlife data for the planet","volume":"47","author":"Ahumada","year":"2020","journal-title":"Environ. Conserv."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"80","DOI":"10.1111\/2041-210X.13099","article-title":"Identifying animal species in camera trap images using deep learning and citizen science","volume":"10","author":"Willi","year":"2019","journal-title":"Methods Ecol. Evol."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"10374","DOI":"10.1002\/ece3.6692","article-title":"Improving the accessibility and transferability of machine learning algorithms for identification of animals in camera trap images: MLWIC2","volume":"10","author":"Tabak","year":"2020","journal-title":"Ecol. Evol."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"165","DOI":"10.1111\/j.1469-1795.2000.tb00241.x","article-title":"Predators and people: Using human densities to interpret declines of large carnivores","volume":"3","author":"Woodroffe","year":"2000","journal-title":"Anim. Conserv."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"484","DOI":"10.1126\/science.abj4216","article-title":"Human-wildlife conflict under climate change","volume":"373","author":"Abrahms","year":"2021","journal-title":"Science"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"224","DOI":"10.1038\/s41558-023-01608-5","article-title":"Climate change as a global amplifier of human\u2013wildlife conflict","volume":"13","author":"Abrahms","year":"2023","journal-title":"Nat. Clim. Change"},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"103157","DOI":"10.1016\/j.ecoinf.2025.103157","article-title":"Automated near real-time monitoring in ecology: Status quo and ways forward","volume":"85","author":"Davison","year":"2025","journal-title":"Ecol. Inform."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"867","DOI":"10.1111\/2041-210X.14036","article-title":"Real-time alerts from AI-enabled camera traps using the Iridium satellite network: A case-study in Gabon, Central Africa","volume":"14","author":"Whytock","year":"2023","journal-title":"Methods Ecol. Evol."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"102527","DOI":"10.1016\/j.ecoinf.2024.102527","article-title":"A systematic study on transfer learning: Automatically identifying empty camera trap images using deep convolutional neural networks","volume":"80","author":"Yang","year":"2024","journal-title":"Ecol. Inform."},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Cunha, F., dos Santos, E.M., Barreto, R., and Colonna, J.G. (2021). Filtering empty camera trap images in embedded systems. 2021 IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), IEEE.","DOI":"10.1109\/CVPRW53098.2021.00276"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"150","DOI":"10.1111\/2041-210X.13504","article-title":"A deep active learning system for species identification and counting in camera trap images","volume":"12","author":"Norouzzadeh","year":"2021","journal-title":"Methods Ecol. Evol."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"287","DOI":"10.1007\/s40264-024-01499-1","article-title":"Performance and Reproducibility of Large Language Models in Named Entity Recognition: Considerations for the Use in Controlled Environments","volume":"48","author":"Dietrich","year":"2025","journal-title":"Drug Saf."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"106","DOI":"10.1038\/s41746-024-01079-8","article-title":"A Critical Assessment of Using ChatGPT for Extracting Structured Data from Clinical Notes","volume":"7","author":"Huang","year":"2024","journal-title":"npj Digit. Med."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"115804","DOI":"10.1016\/j.jbusres.2025.115804","article-title":"Jagged Competencies: Measuring the Reliability of Generative AI in Academic Research","volume":"203","author":"Thomas","year":"2026","journal-title":"J. Bus. Res."},{"key":"ref_18","unstructured":"Venugopal, S., Gazzetti, M., Gkoufas, Y., and Katrinis, K. (2018). Shadow Puppets: Cloud-level Accurate AI Inference at the Speed and Economy of Edge. USENIX Workshop on Hot Topics in Edge Computing (HotEdge), USENIX Association."},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016). You Only Look Once: Unified, Real-Time Object Detection. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE.","DOI":"10.1109\/CVPR.2016.91"},{"key":"ref_20","unstructured":"Bochkovskiy, A., Wang, C.Y., and Liao, H.Y.M. (2020). YOLOv4: Optimal Speed and Accuracy of Object Detection. arXiv."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.Y., and Berg, A.C. (2016). Ssd: Single shot multibox detector. Proceedings of the European Conference on Computer Vision, Springer.","DOI":"10.1007\/978-3-319-46448-0_2"},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Tan, M., Pang, R., and Le, Q.V. (2020). Efficientdet: Scalable and efficient object detection. 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE.","DOI":"10.1109\/CVPR42600.2020.01079"},{"key":"ref_23","first-page":"4061","article-title":"Neural Operator: Learning Maps Between Function Spaces with Applications to PDEs","volume":"24","author":"Kovachki","year":"2023","journal-title":"J. Mach. Learn. Res."},{"key":"ref_24","first-page":"51","article-title":"Fourier Neural Operator Surrogate of Lithium-Ion Battery Models","volume":"7","author":"Viswanathan","year":"2026","journal-title":"J. Mach. Learn. Model. Comput."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"677","DOI":"10.1109\/TPAMI.2016.2599174","article-title":"Long-term recurrent convolutional networks for visual recognition and description","volume":"39","author":"Donahue","year":"2017","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Moschitti, A., Pang, B., and Daelemans, W. (2014). Learning Phrase Representations using RNN Encoder\u2013Decoder for Statistical Machine Translation. Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), Association for Computational Linguistics.","DOI":"10.3115\/v1\/D14-1"},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"102453","DOI":"10.1016\/j.ecoinf.2023.102453","article-title":"Object classification and visualization with edge artificial intelligence for a customized camera trap platform","volume":"79","author":"Nazir","year":"2024","journal-title":"Ecol. Inform."},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"102815","DOI":"10.1016\/j.ecoinf.2024.102815","article-title":"Reliable and efficient integration of AI into camera traps for smart wildlife monitoring based on continual learning","volume":"83","author":"Sanglas","year":"2024","journal-title":"Ecol. Inform."},{"key":"ref_29","doi-asserted-by":"crossref","unstructured":"Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., and Batra, D. (2017). Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization. 2017 IEEE International Conference on Computer Vision (ICCV), IEEE.","DOI":"10.1109\/ICCV.2017.74"},{"key":"ref_30","unstructured":"Jocher, G., Chaurasia, A., and Qiu, J. (2025, December 24). Ultralytics YOLOv8. GitHub Repository, 2023. Version 8.0.0. Available online: https:\/\/github.com\/ultralytics\/ultralytics."},{"key":"ref_31","unstructured":"Tan, M., and Le, Q.V. (2019, January 9\u201315). EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. Proceedings of the 36th International Conference on Machine Learning (ICML), Long Beach, CA, USA."},{"key":"ref_32","doi-asserted-by":"crossref","unstructured":"Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., and Fei-Fei, L. (2009). ImageNet: A Large-Scale Hierarchical Image Database. 2009 IEEE Conference on Computer Vision and Pattern Recognition, IEEE.","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"100322","DOI":"10.1016\/j.fraope.2025.100322","article-title":"Architecture review: Two-stage and one-stage object detection","volume":"12","author":"Mohammed","year":"2025","journal-title":"Frankl. Open"},{"key":"ref_34","unstructured":"(2026, February 17). Flutter Is an Open Source Framework for Building Beautiful, Natively Compiled, Multi-Platform Applications from a Single Codebase. Available online: https:\/\/flutter.dev\/."},{"key":"ref_35","first-page":"304","article-title":"Review of methods to reduce livestock depredation II. Aversive conditioning, deterrents and repellents","volume":"50","author":"Smith","year":"2000","journal-title":"Acta Agric. Scand. Sect. A-Anim. Sci."},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"371","DOI":"10.1111\/acv.12104","article-title":"Effectiveness of low-cost deterrents in decreasing livestock predation by felids: A case in C entral M exico","volume":"17","year":"2014","journal-title":"Anim. Conserv."},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"1241484","DOI":"10.1126\/science.1241484","article-title":"Status and ecological effects of the world\u2019s largest carnivores","volume":"343","author":"Ripple","year":"2014","journal-title":"Science"},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"577","DOI":"10.1007\/s00442-018-4315-z","article-title":"Pumas as ecosystem engineers: Ungulate carcasses support beetle assemblages in the Greater Yellowstone Ecosystem","volume":"189","author":"Barry","year":"2019","journal-title":"Oecologia"},{"key":"ref_39","unstructured":"Elbroch, M. (2020). The Cougar Conundrum: Sharing the World with a Successful Predator, Island Press."},{"key":"ref_40","doi-asserted-by":"crossref","first-page":"110106","DOI":"10.1016\/j.biocon.2023.110106","article-title":"Perspective: Why might removing carnivores maintain or increase risks for domestic animals?","volume":"283","author":"Elbroch","year":"2023","journal-title":"Biol. Conserv."},{"key":"ref_41","doi-asserted-by":"crossref","unstructured":"Cain, J.W., Kay, J.H., Liley, S.G., and Gedir, J.V. (2024). Mule deer (Odocoileus hemionus) Resour. Sel. Trade-Offs Forage Predation Risk. Front. Ecol. Evol., 12.","DOI":"10.3389\/fevo.2024.1121439"},{"key":"ref_42","unstructured":"Sapkota, R., and Karkee, M. (2025). Ultralytics YOLO evolution: An overview of YOLO26, YOLO11, YOLOv8 and YOLOv5 object detectors for computer vision and pattern recognition. arXiv."},{"key":"ref_43","doi-asserted-by":"crossref","unstructured":"Liu, Z., Mao, H., Wu, C.Y., Feichtenhofer, C., Darrell, T., and Xie, S. (2022, January 18\u201324). A convnet for the 2020s. Proceedings of the 2022 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, USA.","DOI":"10.1109\/CVPR52688.2022.01167"},{"key":"ref_44","unstructured":"(2026, February 17). Pumaguard v21. Available online: https:\/\/github.com\/PEEC-Nature-Youth-Group\/pumaguard\/releases\/tag\/v21."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/4\/1366\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,23]],"date-time":"2026-02-23T09:03:37Z","timestamp":1771837417000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/4\/1366"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,21]]},"references-count":44,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2026,2]]}},"alternative-id":["s26041366"],"URL":"https:\/\/doi.org\/10.3390\/s26041366","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,21]]}}}