Deep Learning for Detecting Abandoned Dogs

Oluwakemi Akinwehinmi; Alberto Tena; Francisco Javier Mora; Francesc Solsona; Pedro Arnau del Amo

doi:10.9781/ijimai.2026.2222

Authors

Oluwakemi Akinwehinmi Universitat de Lleida https://orcid.org/0009-0005-9227-6630
Alberto Tena Universitat de Lleida https://orcid.org/0000-0002-3999-5549
Francisco Javier Mora International Center for Numerical Methods in Engineering https://orcid.org/0000-0001-7142-2559
Francesc Solsona Universitat de Lleida https://orcid.org/0000-0002-4830-9184
Pedro Arnau del Amo Findspo (Spain) https://orcid.org/0000-0002-3718-8559

DOI:

https://doi.org/10.9781/ijimai.2026.2222

Keywords:

Activity Recognition, Algorithms, Anomaly Detection, Deep Learning, Edge Computing, Machine Learning, Machine Vision

Supporting Agencies

This work has been supported by the Generalitat de Catalunya for the financial support to the primary author, beneficiary of a pre-doctoral grant funded under the Program contract between the Administration of the Generalitat of Catalonia, through the Department of Territory and Sustainability and the Department of Business and Knowledge, and the International Center for Numerical Methods in Engineering (CIMNE), for the period 2020-2023. This research was also been developed within the PIKSEL project, “Portal for the integration of knowledge for a sustainable ecosystems and land management” funded by Generalitat de Catalunya, through the Department of Territory and Sustainability and the Department of Climate Action. The authors also acknowledge the financial support through the Severo Ochoa Centers of Excellence Program (CEX 2018- 000797-S) funded by MCIN/AEI/10.13039/501100011033

Abstract

This research paper presents a methodology consisting of an algorithm and a workflow for finding abandoned dogs in natural surroundings. We propose a temporal-contextual methodology for identifying abandoned pets in public areas. This involves employing a temporal rule alongside context-sensitive object identification, whereby dog bounding boxes are deliberately expanded to encompass proximate visual indicators suggestive of abandonment. The proposed approach uses object detection techniques, trajectory analysis, and image segmentation to quickly differentiate between abandoned and owned dogs. The research addresses key challenges, such as data scarcity and the complexity of distinguishing between abandoned and accompanied dogs. To address the issue of sufficient and adequate training corpus to identify an abandoned dog, the research article employs single-channel image augmentation methods that improve model recall and precision by 4%. Several object detection algorithms were evaluated, and our findings indicate that single-stage detectors like YOLOv8 achieved a better trade-off between classification performance and speed for detecting abandoned dogs compared to multi-stage detectors like Faster R-CNN, reaching a mean average precision up to 86% and an inference time of 0.3 ms per frame. This research study contributes to animal welfare, biodiversity conservation, and public safety by providing a scalable solution for monitoring abandoned animals in diverse environments. The findings demonstrate the impact of object detection techniques on improving the generalization of deep learning models for real-world applications.

Downloads

Download data is not yet available.

Author Biographies

Oluwakemi Akinwehinmi, Universitat de Lleida

Oluwakemi Akinwehinmi is currently pursuing a Doctor of Philosophy (PhD) in Computer Science at Universitat de Lleida, Spain (2022–2025). She holds a Master of Science and a Bachelor of Science in Computer Science from the University of Ibadan, Nigeria (2018–2020 and 2012–2016, respectively). Her research interests include artificial intelligence, deep learning, computer vision, generative artificial intelligence, and cloud computing. She is a member and volunteer of Data Science Nigeria, Data Science Africa, and Women in Machine Learning and Artificial Intelligence.

Alberto Tena, Universitat de Lleida

Alberto Tena is a lecturer professor at the University of Lleida (UdL), affiliated with the Department of Computer Science and Digital Design, where he specializes in distributed computing. He received a Ph.D. in Engineering and Information Technologies from the University of Lleida in 2022, and holds a Bachelor’s degree in Telecommunications Engineering and a Master’s degree in Applied Telecommunications and Engineering Management, both obtained in 2009 from the Polytechnic University of Catalonia (UPC), Spain. His research focuses on distributed monitoring systems, combining digital signal processing, distributed computing architectures, and machine learning applied to multimodal data.

Francisco Javier Mora, International Center for Numerical Methods in Engineering

Francisco Javier Mora received a degree in Telecommunications Engineering from the Polytechnic University of Catalonia (UPC) in 1992 and a Ph.D. in Electronic Engineering in 1998. Since 1998, he has been a Senior Researcher and Project Manager at the International Center for Numerical Methods in Engineering (CIMNE). His early career focused on computational electromagnetics and finite element simulation. Currently, he specializes in the digitalization of the AECO industry, focusing on Building Information Modeling (BIM) and expanded reality (XR) applications. His research integrates AR/VR for quality control, monitoring, and highprecision positioning. He plays a leading role in construction tech, bridging the gap between numerical methods and immersive digital tools for industrial innovation.

Francesc Solsona, Universitat de Lleida

Francesc Solsona is a full professor at the Universitat de Lleida. His research experience is reflected in four awarded Research Six-Year Terms (sexenios), an h-index of 21 (Google Scholar), and over 1,700 citations. He has published more than 130 journal articles and book chapters, many indexed in the JCR, and has contributed to over 100 national and international conferences. His research focuses on cluster and distributed computing, cloud, fog/edge, and IoT systems, with particular emphasis on the design of algorithms, models, and simulation environments for task scheduling in parallel and distributed architectures. His work is highly interdisciplinary, spanning computer science, medical informatics, epidemiology, artificial intelligence, signal processing, and operations research. He is a member of the Distributed Computing Group (GCD) at the University of Lleida, a consolidated research group accredited by the Government of Catalonia. He has supervised 10 PhD theses, several with international mention and excellence awards, and has participated in more than 30 competitive research and technology transfer projects, including national, European, and industrial initiatives.

Pedro Arnau del Amo, Findspo (Spain)

Pedro Arnau del Amo holds a Ph.D. in Physical Oceanography from the Polytechnic University of Catalonia (UPC), Spain. His doctoral thesis on mesoscale marine circulation in the Catalan Sea was awarded First Prize in the Sustainable Development Thesis Competition by the AGBAR Foundation. He currently serves as Chief Climate Resilience Scientist at Findspo. He has led national and international research projects as principal investigator and has been involved in initiatives focused on protecting ecosystem services and integrating information and communication technologies (ICT) into environmental monitoring. His expertise includes remote sensing, geographic information systems (GIS), and machine learning applied to environmental and marine sciences.

References

[1] W. Hu, T. Tan, L. Wang, S. Maybank, “A survey on visual surveillance of object motion and behaviors,” IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 34, no. 3, pp. 334–352, 2004, doi: https://doi.org/10.1109/TSMCC.2004.829274.

[2] J. Redmon, A. Farhadi, “Yolov3: An incremental improvement,” arXiv preprint arXiv:1804.02767, 2018, doi: https://doi.org/10.48550/arXiv.1804.02767.

[3] J. Hughes, D. W. Macdonald, “A review of the interactions between freeroaming domestic dogs and wildlife,” Biological Conservation, vol. 157, pp. 341–351, 2013, doi: https://doi.org/10.1016/j.biocon.2012.07.005.

[4] World Health Organization, “Rabies,” 2023. Accessed: May 28, 2025, [Online]. Available: https://www.who.int/news-room/factsheets/detail/rabies.

[5] K. Hampson, L. Coudeville, T. Lembo, M. Sambo, A. Kieffer, M. Attlan, J. Barrat, J. D. Blanton, D. J. Briggs, S. Cleaveland, P. Costa, C. M. Freuling, E. Hiby, L. Knopf, F. Leanes, F.-X. Meslin, A. Metlin, M. E. Miranda, T. Müller, L. H. Nel, S. Recuenco, C. E. Rupprecht, C. Schumacher, L. Taylor, M. A. N. Vigilato, J. Zinsstag, J. Dushoff, “Estimating the global burden of endemic canine rabies,” PLoS Neglected Tropical Diseases, vol. 9, no. 4, p. e0003709, 2015, doi: https://doi.org/10.1371/journal.pntd.0003709.

[6] G. Jocher, “Yolov5 by ultralytics,” 2020. doi: https://doi.org/10.5281/zenodo.3908559.

[7] C. Li, L. Li, H. Jiang, K. Weng, Y. Geng, L. Li, Z. Ke, Q. Li, M. Cheng, W. Nie, Y. Li, B. Zhang, Y. Liang, L. Zhou, X. Xu, X. Chu, X. Wei, X. Wei, “Yolov6: A single-stage object detection framework for industrial applications,” arXiv preprint arXiv:2209.02976, 2022, doi: https://doi.org/10.48550/arXiv.2209.02976.

[8] R. Girshick, “Fast r-cnn,” in Proceedings of the 2015 IEEE International Conference on Computer Vision (ICCV), Santiago, Chile, 2015, pp. 1440–1448.

[9] S. Ren, K. He, R. Girshick, J. Sun, “Faster rcnn: Towards real-time object detection with region proposal networks,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137–1149, 2017, doi: https://doi.org/10.1109/TPAMI.2016.2577031.

[10] R. Girshick, I. Radosavovic, G. Gkioxari, P. Dollár, K. He, “Detectron,” 2018. Accessed: May 28, 2025, [Online]. Available: https://github.com/facebookresearch/detectron.

[11] Y. Wu, A. Kirillov, F. Massa, W.-Y. Lo, R. Girshick, “Detectron2,” 2019. Accessed: May 28, 2025, [Online]. Available: https://github.com/facebookresearch/detectron2.

[12] L. Ferrando, G. Gera, C. Regazzoni, “Classification of unattended and stolen objects in videosurveillance systems,” in Proceedings of the 2006 IEEE International Conference on Video and Signal Based Surveillance, Sydney, Australia, 2006, pp. 21– 21.

[13] F. I. Bashir, A. A. Khokhar, D. Schonfeld, “Realtime motion trajectorybased indexing and retrieval of video sequences,” IEEE Transactions on Multimedia, vol. 9, no. 1, pp. 58–65, 2007, doi: https://doi.org/10.1109/TMM.2006.886346.

[14] L. Taylor, G. Nitschke, “Improving deep learning using generic data augmentation,” arXiv preprint arXiv:1708.06020, 2017, doi: https://doi.org/10.48550/arXiv.1708.06020.

[15] C. Shorten, T. M. Khoshgoftaar, “A survey on image data augmentation for deep learning,” Journal of Big Data, vol. 6, no. 1, p. 60, 2019, doi: https://doi.org/10.1186/s40537-019-0197-0.

[16] CIMNE, “PIKSEL: Portal for the Integration of Knowledge for Sustainable Ecosystems and Land Management,” 2024. Accessed: May 28, 2025, [Online]. Available: https://pikselweb.cimne.com/.

[17] B. Benjdira, T. Khursheed, A. Koubaa, A. Ammar, K. Ouni, “Car detection using unmanned aerial vehicles: Comparison between faster r-cnn and yolov3,” arXiv preprint arXiv:1812.10968, 2018, doi: https://doi.org/10.48550/arXiv.1812.10968.

[18] E. D. Cubuk, B. Zoph, D. Mane, V. Vasudevan, Q. V. Le, “Autoaugment: Learning augmentation policies from data,” arXiv preprint arXiv:1805.09501, 2019, doi: https://doi.org/10.48550/arXiv.1805.09501.

[19] S. Lim, I. Kim, T. Kim, C. Kim, S. Kim, “Fast autoaugment,” in Proceedings of the 33rd International Conference on Neural Information Processing Systems, Vancouver, BC, Canada, 2019.

[20] G. Gibbins, G. Newsam, M. Brooks, “Detecting suspicious background changes in video surveillance of busy scenes,” in Proceedings of the Third IEEE Workshop on Applications of Computer Vision (WACV’96), Sarasota, FL, USA, 1996, pp. 22–26.

[21] A. H. Freedman, I. Gronau, R. M. Schweizer, D. Ortega-Del Vecchyo, E. Han, P. M. Silva, et al., “Genome sequencing highlights the dynamic early history of dogs,” PLoS Genetics, vol. 10, no. 1, p. e1004016, 2014, doi: https://doi.org/10.1371/journal.pgen.1004016.

[22] M. E. Gompper, “The dog–human–wildlife interface: Assessing the scope of the problem,” in Free-Ranging Dogs and Wildlife Conservation, Oxford University Press, 2013.

[23] S. Barnard, C. Ippoliti, D. Di Flaviano, A. De Ruvo, S. Messori, A. Giovannini, P. Dalla Villa, “Smartphone and gps technology for freeroaming dog population surveillance - a methodological study,” Veterinaria Italiana, vol. 51, no. 3, pp. 165–172, 2015, doi: https://doi.org/10.12834/VetIt.233.2163.3.

[24] P. Howell, At Home and Astray: The Domestic Dog in Victorian Britain. Charlottesville, VA, USA: University of Virginia Press, 2015.

[25] O. Akinwehinmi, P. Arnau, F. Solsona, A. Priegue, J. Jimenez, “Paws for concern: Spotting strays and abandoned pets in natural spaces,” in 2023 IEEE International Smart Cities Conference (ISC2), Bucharest, Romania, 2023, pp. 1–7.

[26] J. Ruiz-Chavez, J. Salvador-Meneses, C. MejíaAstudillo, S. DiazQuilachamin, “Analysis of dogs’ abandonment problem using georeferenced multiagent systems,” in From Bioinspired Systems and Biomedical Applications to Machine Learning, Cham, Switzerland: Springer International Publishing, 2019, pp. 297–306.

[27] B. Graham, “Fractional max-pooling,” arXiv preprint arXiv:1412.6071, 2015, doi: https://doi.org/10.48550/arXiv.1412.6071.

[28] Ruman, “Yolo data augmentation explained: Turbocharge your object detection model,” 2023. Accessed: May 28, 2025, [Online]. Available: https://www.rumn.ai/blog/yolodata-augmentation-explained.

[29] J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, F.F. Li, “Imagenet: A large-scale hierarchical image database,” in Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Miami, FL, USA, 2009, pp. 248–255.

[30] O. Akinwehinmi, P. Arnau, F. Solsona, “Improving abandoned pet detection in natural spaces through adaptive single-coloured image augmentation techniques,” in Proceedings of the 15th Congress of the International Colour Association (AIC 2023), Chiang Rai, Thailand, 2023.

[31] T. DeVries, G. W. Taylor, “Improved regularization of convolutional neural networks with cutout,” arXiv preprint arXiv:1708.04552, 2017, doi: https://doi.org/10.48550/arXiv.1708.04552.

[32] H. Zhang, M. Cisse, Y. N. Dauphin, D. Lopez-Paz, “Mixup: Beyond empirical risk minimization,” arXiv preprint arXiv:1710.09412, 2018, doi: https://doi.org/10.48550/arXiv.1710.09412.

[33] Y. Guo, Y. Mao, R. Zhang, “Mixup as locally linear out-of-manifold regularization,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 33, Honolulu, HI, USA, 2019, pp. 3714–3722.

[34] S. Yun, D. Han, S. Chun, S. J. Oh, Y. Yoo, J. Choe, “Cutmix: Regularization strategy to train strong classifiers with localizable features,” in Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, South Korea, 2019, pp. 6022–6031.

[35] A. Bruno, E. Ardizzone, S. Vitabile, M. Midiri, “A novel solution based on scale invariant feature transform descriptors and deep learning for the detection of suspicious regions in mammogram images,” Journal of Medical Signals and Sensors, vol. 10, no. 3, pp. 158–173, 2020, doi: https://doi.org/10.4103/jmss.JMSS_31_19.

[36] B. Recht, R. Roelofs, L. Schmidt, V. Shankar, “Do cifar-10 classifiers generalize to cifar10?,” arXiv preprint arXiv:1806.00451, 2018, doi: https://doi.org/10.48550/arXiv.1806.00451.

[37] A. Jurio, M. Pagola, “ M. Galar, C. LopezMolina, D. Paternain, A comparison study of different color spaces in clustering based image segmentation,” in Information Processing and Management of Uncertainty in Knowledge-Based Systems. Applications, Berlin, Heidelberg: Springer Berlin Heidelberg, 2010, pp. 532–541.

[38] Y. Bengio, F. Bastien, A. Bergeron, N. Boulanger–Lewandowski, T. Breuel, Y. Chherawala, M. Cisse, M. Côté, D. Erhan, J. Eustache, X. Glorot, X. Muller, S. Pannetier Lebeuf, R. Pascanu, S. Rifai, F. Savard, G.

Sicard, “Deep learners benefit more from out-of-distribution examples,” in Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics (AISTATS), vol. 15 of Proceedings of Machine Learning Research, Fort Lauderdale, FL, USA, 2011, pp. 164–172.

[39] E. Azizi, L. Zaman, “Deep learning pet identification using face and body,” Information, vol. 14, no. 5, 2023, doi: https://doi.org/10.3390/info14050278.

[40] P. Kasnesis, V. Doulgerakis, D. Uzunidis, D. G. Kogias, S. I. Funcia, M. B. González, C. Giannousis, C. Z. Patrikakis, “Deep learning empowered wearable-based behavior recognition for search and rescue dogs,” Sensors, vol. 22, no. 3, 2022, doi: https://doi.org/10.3390/s22030993.

[41] Y. Sangve, Y. Firke, S. Shinde, S. Patil, P. Shinde, P. Mitake, “A comprehensive survey on real-time animal (dog) detection system using artificial intelligence methods,” in Proceedings of the 4th International Conference on Artificial Intelligence and Smart Energy, Cham, Switzerland: Springer Nature Switzerland, 2024, pp. 260–275.

[42] O. Akinwehinmi, “Abandoned dogs detection dataset,” 2025. Accessed: Jun. 3, 2025, [Online]. Available: https://www.kaggle. com/datasets/oluwakemiakinwehinmi/ abandoned-dogs-detection.

[43] O. Akinwehinmi, “Abandoned dog detection - code for reproduction,” 2025. Accessed: Jun. 3, 2025, [Online]. Available: https://github.com/DrKem/Abandoned-DogDetection/blob/main/Codes

[44] B. D. Lucas, T. Kanade, “An iterative image registration technique with an application to stereo vision,” in Proceedings of the 7th International Joint Conference on Artificial Intelligence (IJCAI’81), San Francisco, CA, USA, 1981, pp. 674–679.

[45] B. K. Horn, B. G. Schunck, “Determining optical flow,” Artificial Intelligence, vol. 17, no. 1, pp. 185– 203, 1981, doi: https://doi.org/10.1016/00043702(81)90024-2.

[46] A. Bewley, Z. Ge, L. Ott, F. Ramos, B. Upcroft, “Simple online and realtime tracking,” in Proceedings of the 2016 IEEE International Conference on Image Processing (ICIP), Phoenix, AZ, USA, 2016, pp. 3464–3468.

[47] N. Wojke, A. Bewley, D. Paulus, “Simple online and realtime tracking with a deep association metric,” in Proceedings of the 2017 IEEE International Conference on Image Processing (ICIP), Beijing, China, 2017,pp. 3645–3649.

[48] K. Chatfield, A. Zisserman, “Visor: Towards onthe-fly large-scale object category retrieval,” in Computer Vision – ACCV 2012, Berlin, Heidelberg: Springer Berlin Heidelberg, 2013, pp. 432–446.

[49] M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, A. Zisserman, “The pascal visual object classes (voc) challenge,” International Journal of Computer Vision, vol. 88, no. 2, pp. 303–338, 2010, doi: https://doi.org/10.1007/s11263-009-0275-4.

Deep Learning for Detecting Abandoned Dogs

Authors

DOI:

Keywords: