{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,7]],"date-time":"2026-10-07T13:12:38Z","timestamp":1791378758416,"version":"4.3.2"},"reference-count":42,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,2,18]],"date-time":"2026-02-18T00:00:00Z","timestamp":1771372800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100012331","name":"Agentschap Innoveren en Ondernemen","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100012331","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Toxicol."],"abstract":"<jats:p>Behavioral monitoring of laboratory animals is essential for evaluating drug safety, yet existing assessments are typically limited to in-room observations by technicians. Here, we introduce our versatile AI model pipeline, composed of interconnected artificial neural networks that leverage end-to-end learning based solely on video-derived appearance features of canines. This non-invasive approach enables detailed mapping of activity, behavior and clinical signs at individual animal level under diverse conditions. To validate its real-world application, we conducted extensive field testing on hours of footage. Trained on a large, annotated dataset, our model can accurately multi-track up to three group-housed canines using color-coded reflective harnesses, achieving high re-identification accuracies (\u226592.5%) and IDF1 scores up to 99.9%. AI-derived locomotor activity showed a strong correlation with accelerometer-based measurements (r = 0.965). Our AI model detects 11 behavior and clinical observation classes, with a mean class accuracy of 48% and individual accuracies up to 93%. As such, a detailed time-specific quantitative output is available for activity, mobility, pose, eating, drinking and specific clinical signs (ataxia, anxiety, circling, convulsions, head shaking, involuntary muscle movements, limping, limb stiff, vomiting). Our innovative approach brings holistic behavioral and health monitoring in canines closer to routine practice and contributes towards the 3Rs principles.<\/jats:p>","DOI":"10.3389\/ftox.2026.1758963","type":"journal-article","created":{"date-parts":[[2026,2,18]],"date-time":"2026-02-18T06:56:10Z","timestamp":1771397770000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":3,"title":["Appearance-based computer vision pipeline for multi-animal monitoring of canine activity, behavior and clinical observations"],"prefix":"10.3389","volume":"8","author":[{"given":"Eline","family":"Eberhardt","sequence":"first","affiliation":[{"name":"Non-Clinical Safety and Submissions, Preclinical Sciences and Translational Safety, J&J Innovative Medicine, Janssen Research & Development","place":["Beerse, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jef","family":"Plochaet","sequence":"additional","affiliation":[{"name":"Eavise- PSI, department of Electrical Engineering ESAT, KU Leuven","place":["Sint-Katelijne-Waver, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tanguy","family":"Ophoff","sequence":"additional","affiliation":[{"name":"Eavise- PSI, department of Electrical Engineering ESAT, KU Leuven","place":["Sint-Katelijne-Waver, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Floris","family":"De Feyter","sequence":"additional","affiliation":[{"name":"Eavise- PSI, department of Electrical Engineering ESAT, KU Leuven","place":["Sint-Katelijne-Waver, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sarah","family":"De Landtsheer","sequence":"additional","affiliation":[{"name":"Scientific and In vivo Strategies, Preclinical Sciences and Translational Safety, J&J Innovative Medicine, Janssen Research & Development","place":["Beerse, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Greet","family":"Teuns","sequence":"additional","affiliation":[{"name":"Global Safety Pharmacology, Preclinical Sciences and Translational Safety, J&J Innovative Medicine, Janssen Research & Development","place":["Beerse, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maarten","family":"Vergauwen","sequence":"additional","affiliation":[{"name":"Scientific and In vivo Strategies, Preclinical Sciences and Translational Safety, J&J Innovative Medicine, Janssen Research & Development","place":["Beerse, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bianca","family":"Feyen","sequence":"additional","affiliation":[{"name":"Non-Clinical Safety and Submissions, Preclinical Sciences and Translational Safety, J&J Innovative Medicine, Janssen Research & Development","place":["Beerse, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Toon","family":"Goedem\u00e9","sequence":"additional","affiliation":[{"name":"Eavise- PSI, department of Electrical Engineering ESAT, KU Leuven","place":["Sint-Katelijne-Waver, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ivan","family":"Kopljar","sequence":"additional","affiliation":[{"name":"Non-Clinical Safety and Submissions, Preclinical Sciences and Translational Safety, J&J Innovative Medicine, Janssen Research & Development","place":["Beerse, Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,2,18]]},"reference":[{"key":"B1","first-page":"1","article-title":"Comprehensive evaluation of deep learning based detection methods for vehicle detection in aerial imagery","author":"Acatay","year":"2018"},{"key":"B2","first-page":"6836","article-title":"Vivit: a video vision transformer","author":"Arnab","year":"2021"},{"key":"B3","doi-asserted-by":"publisher","first-page":"52","DOI":"10.1016\/j.vascn.2019.03.004","article-title":"EEG: characteristics of drug-induced seizures in rats, dogs and non-human primates","volume":"97","author":"Authier","year":"2019","journal-title":"J. Pharmacol. Toxicol. Methods"},{"key":"B4","doi-asserted-by":"publisher","first-page":"4560","DOI":"10.1038\/s41467-020-18441-5","article-title":"Automated markerless pose estimation in freely moving macaques with OpenMonkeyStudio","volume":"11","author":"Bala","year":"2020","journal-title":"Nat. Commun."},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.1098\/rsif.2014.0672","article-title":"Mapping the stereotyped behaviour of freely moving fruit flies","volume":"11","author":"Berman","year":"2014","journal-title":"J. R. Soc. Interface"},{"key":"B6","doi-asserted-by":"publisher","first-page":"6","DOI":"10.3389\/ftox.2024.1377542","article-title":"Digitalization of toxicology: improving preclinical to clinical translation","volume":"6","author":"Berridge","year":"2024","journal-title":"Front. Toxicol."},{"key":"B7","article-title":"Yolov4: optimal speed and accuracy of object detection","author":"Bochkovskiy","year":"2020"},{"key":"B8","doi-asserted-by":"publisher","first-page":"e63377","DOI":"10.7554\/eLife.63377","article-title":"DeepEthogram, a machine learning pipeline for supervised behavior classification from raw pixels","volume":"10","author":"Bohnslav","year":"2021","journal-title":"Elife"},{"key":"B9","first-page":"558","article-title":"Tide: a general toolbox for identifying object detection errors","author":"Bolya","year":"2020"},{"key":"B10","first-page":"13052","article-title":"Mammalnet: a large-scale video benchmark for mammal recognition and behavior understanding","author":"Chen","year":"2023"},{"key":"B11","doi-asserted-by":"publisher","first-page":"106896","DOI":"10.1016\/j.vascn.2020.106896","article-title":"Use of the actiwatch-mini\u00ae in dog safety studies as an early indicator for drug-induced behavioural changes","volume":"104","author":"Cools","year":"2020","journal-title":"J. Pharmacol. Toxicol. Methods"},{"key":"B12","doi-asserted-by":"publisher","first-page":"1679","DOI":"10.1109\/TAES.2016.140952","article-title":"On implementing 2D rectangular assignment algorithms","volume":"52","author":"Crouse","year":"2016","journal-title":"IEEE Trans. Aerosp. Electron. Syst."},{"key":"B13","first-page":"702","article-title":"Randaugment: practical automated data augmentation with a reduced search space","author":"Cubuk","year":"2020"},{"key":"B14","article-title":"An image is worth 16x16 words: transformers for image recognition at scale","author":"Dosovitskiy","year":"2020"},{"key":"B15","doi-asserted-by":"publisher","first-page":"e0326916","DOI":"10.1371\/journal.pone.0326916","article-title":"Application of video surveillance in preclinical safety studies in canines: understanding the interobserver reliability and validity to recognize clinical behavior","volume":"20","author":"Eberhardt","year":"2025","journal-title":"PLOS ONE"},{"key":"B16","doi-asserted-by":"publisher","first-page":"303","DOI":"10.1007\/s11263-009-0275-4","article-title":"The pascal visual object classes (VOC) challenge","volume":"88","author":"Everingham","year":"2010","journal-title":"Int. J. Comput. Vis."},{"key":"B17","doi-asserted-by":"publisher","first-page":"17669","DOI":"10.1007\/s00521-024-10042-3","article-title":"Advanced techniques for automated emotion recognition in dogs from video data through deep learning","volume":"36","author":"Franzoni","year":"2024","journal-title":"Neural Comput. Appl."},{"key":"B18","doi-asserted-by":"publisher","first-page":"109961","DOI":"10.1016\/j.compag.2025.109961","article-title":"Occlusion-robust detection of sow-induced piglet crushing incidents using spatial and motion reasoning","volume":"231","author":"Gan","year":"2025","journal-title":"Comput. Electron. Agric."},{"key":"B19","first-page":"770","article-title":"Deep residual learning for image recognition","author":"He","year":"2016"},{"key":"B20","article-title":"Motmetrics: a python library for multi-object tracking metrics","author":"Heindl","year":"2019","journal-title":"GitHub"},{"key":"B21","doi-asserted-by":"publisher","first-page":"17","DOI":"10.3389\/fnins.2023.1149027","article-title":"Open-source software for automated rodent behavioral analysis","volume":"17","author":"Isik","year":"2023","journal-title":"Front. Neurosci."},{"key":"B22","doi-asserted-by":"publisher","first-page":"325","DOI":"10.1007\/bf02278710","article-title":"A shortest augmenting path algorithm for dense and sparse linear assignment problems","volume":"38","author":"Jonker","year":"1987","journal-title":"Computing"},{"key":"B23","doi-asserted-by":"publisher","first-page":"256","DOI":"10.1186\/s12915-023-01751-7","article-title":"A systematic review of the development and application of home cage monitoring in laboratory mice and rats","volume":"21","author":"Kahnau","year":"2023","journal-title":"BMC Biol."},{"key":"B24","doi-asserted-by":"publisher","first-page":"35","DOI":"10.1115\/1.3662552","article-title":"A new approach to linear filtering and prediction problems","volume":"82","author":"Kalman","year":"1960","journal-title":"Trans. ASME- J. Basic Eng."},{"key":"B25","doi-asserted-by":"publisher","first-page":"496","DOI":"10.1038\/s41592-022-01443-0","article-title":"Multi-animal pose estimation, identification and tracking with DeepLabCut","volume":"19","author":"Lauer","year":"2022","journal-title":"Nat. Methods"},{"key":"B26","first-page":"268","article-title":"Improving object detection in VHR aerial orthomosaics","author":"Ophoff","year":"2022"},{"key":"B27","doi-asserted-by":"publisher","first-page":"123","DOI":"10.1177\/10915818231218984","article-title":"The spontaneous incidence of neurological clinical signs in preclinical species using cage-side observations or high-definition video monitoring: a retrospective analysis","volume":"43","author":"Orciani","year":"2024","journal-title":"Int. J. Toxicol."},{"key":"B28","doi-asserted-by":"publisher","first-page":"486","DOI":"10.1038\/s41592-022-01426-1","article-title":"SLEAP: a deep learning system for multi-animal pose tracking","volume":"19","author":"Pereira","year":"2022","journal-title":"Nat. Methods"},{"key":"B29","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR.2017.690","article-title":"YOLO9000: better, faster, stronger","author":"Redmon","year":"2017"},{"key":"B30","article-title":"Yolov3: an incremental improvement","author":"Redmon","year":"2018"},{"key":"B31","doi-asserted-by":"publisher","first-page":"17","DOI":"10.48550\/arXiv.1609.01775","article-title":"Performance measures and a data set for multi-target, multi-camera tracking","author":"Ristani","year":"2016","journal-title":"Comput. Vis. \u2013 ECCV 2016 Work."},{"key":"B32","unstructured":"Scipy.optimize: linear_sum_assignment\n          \n          \n          2008"},{"key":"B33","doi-asserted-by":"publisher","first-page":"7","DOI":"10.3389\/ftox.2025.1655330","article-title":"Rodent home cage monitoring for preclinical safety pharmacology assessment: results of a multi-company validation evaluating nonclinical and clinical data from three compounds","volume":"7","author":"Sillito","year":"2025","journal-title":"Front. Toxicol."},{"key":"B34","doi-asserted-by":"publisher","first-page":"10078","DOI":"10.48550\/arXiv.2203.12602","article-title":"Videomae: masked autoencoders are data-efficient learners for self-supervised video pre-training","volume":"35","author":"Tong","year":"2022","journal-title":"Adv. Neural Information Processing Systems"},{"key":"B35","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1805.09512","article-title":"You only look twice: rapid multi-scale object detection in satellite imagery","author":"Van Etten","year":"2018","journal-title":"arXiv Preprint arXiv:1805.09512"},{"key":"B36","unstructured":"Acepromazine\n          \n          \n          \n          2023"},{"key":"B37","doi-asserted-by":"publisher","first-page":"1121","DOI":"10.1016\/j.neuron.2015.11.031","article-title":"Mapping sub-second structure in mouse behavior","volume":"88","author":"Wiltschko","year":"2015","journal-title":"Neuron"},{"key":"B38","doi-asserted-by":"publisher","first-page":"106016","DOI":"10.1016\/j.compag.2021.106016","article-title":"Using a CNN-LSTM for basic behaviors detection of a single dairy cow in a complex environment","volume":"182","author":"Wu","year":"2021","journal-title":"Comput. Electron. Agric."},{"key":"B39","first-page":"6023","article-title":"Cutmix: regularization strategy to train strong classifiers with localizable features","author":"Yun","year":"2019"},{"key":"B40","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1710.09412","article-title":"Mixup: beyond empirical risk minimization","author":"Zhang","year":"2017","journal-title":"arXiv Preprint arXiv:1710.09412"},{"key":"B41","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2505.00569","article-title":"AnimalMotionCLIP: embedding motion in CLIP for animal behavior analysis","author":"Zhong","year":"2025","journal-title":"arXiv Preprint arXiv:2505.00569"},{"key":"B42","doi-asserted-by":"publisher","first-page":"e175824","DOI":"10.1172\/JCI175824","article-title":"FDA modernization act 2.0: transitioning beyond animal models with human cells, organoids, and AI\/ML-based approaches","volume":"133","author":"Zushin","year":"2023","journal-title":"J. Clin. Investigation"}],"container-title":["Frontiers in Toxicology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/ftox.2026.1758963\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,18]],"date-time":"2026-02-18T06:56:13Z","timestamp":1771397773000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/ftox.2026.1758963\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,18]]},"references-count":42,"alternative-id":["10.3389\/ftox.2026.1758963"],"URL":"https:\/\/doi.org\/10.3389\/ftox.2026.1758963","relation":{},"ISSN":["2673-3080"],"issn-type":[{"value":"2673-3080","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,18]]},"article-number":"1758963"}}