Segmentation-enhanced ensemble deep learning for animal species classification

Year 2026, Volume: 6 Issue: 1, 258 - 274, 31.01.2026

Mustafa Ceylan , Andaç Fındıkçı , Mustafa Yasin Erten , Hüseyin Aydilek

https://doi.org/10.61112/jiens.1720934

https://izlik.org/JA29PW49SA

Abstract

In recent years, there has been a significant surge in the applications of artificial intelligence, with remarkable advancements recorded in their predictive and problem-solving capabilities. Image processing technologies have emerged as a pivotal component of this advancement, enabling the development of models capable of detecting fine details with high accuracy and thereby providing critical solutions for areas such as biodiversity conservation. The accurate identification and classification of animal species are of paramount importance for monitoring endangered species, assessing ecosystem health, and planning conservation efforts. However, traditional methods are often time-consuming, costly, and prone to human error, thus heightening the need for more reliable and efficient systems. In this study, a dataset comprising 74 different animal species was utilized. The images in this dataset were first segmented using the SAM 2 model, after which independent models were trained with deep learning architectures such as InceptionV3, Xception, and DenseNet169, and hyperparameter optimization was performed using the Bayesian search method. To further enhance classification performance, a Soft Voting ensemble learning approach was employed, achieving an accuracy rate of approximately 93%. This proposed model has been named The Deep Pet Ensemble. The results indicate that while artificial intelligence and image processing techniques can serve as powerful tools for animal species recognition, they also present a rapid, cost-effective alternative for supporting biodiversity conservation, ecosystem sustainability, and scientific research.

Keywords

Deep learning , Pet Image Classification , Segmentation , Ensemble Learning , Artificial Intelligience

Hayvan Türlerinin Sınıflandırılması için Segmentasyon Destekli Topluluk Derin Öğrenme Yaklaşımı

https://izlik.org/JA29PW49SA

Abstract

Son yıllarda yapay zekâ uygulamalarında önemli bir artış gözlemlenmiş; bu alanda tahmin ve problem çözme yeteneklerinde dikkate değer ilerlemeler kaydedilmiştir. Görüntü işleme teknolojileri, bu gelişimin kilit bileşenlerinden biri olarak öne çıkmakta; yüksek doğrulukla ince detayları tespit edebilen modellerin geliştirilmesini mümkün kılmakta ve biyolojik çeşitliliğin korunması gibi alanlarda kritik çözümler sunmaktadır. Hayvan türlerinin doğru bir şekilde tanımlanması ve sınıflandırılması, nesli tükenmekte olan türlerin izlenmesi, ekosistem sağlığının değerlendirilmesi ve koruma çalışmalarının planlanması açısından büyük önem taşımaktadır. Ancak geleneksel yöntemler genellikle zaman alıcı, maliyetli ve insan hatasına açık olduğundan, daha güvenilir ve verimli sistemlere duyulan ihtiyaç artmaktadır. Bu çalışmada, 74 farklı hayvan türünden oluşan bir veri kümesi kullanılmıştır. Veri kümesindeki görseller öncelikle SAM 2 modeli ile segmentasyona tabi tutulmuş; ardından InceptionV3, Xception ve DenseNet169 gibi derin öğrenme mimarileriyle bağımsız modeller eğitilmiş ve hiperparametre optimizasyonu Bayesyen arama yöntemiyle gerçekleştirilmiştir. Sınıflandırma performansını daha da artırmak amacıyla Soft Voting (Yumuşak Oylama) topluluk öğrenme yaklaşımı uygulanmış ve yaklaşık %93 doğruluk oranı elde edilmiştir. Önerilen bu model "The Deep Pet Ensemble" olarak adlandırılmıştır. Sonuçlar, yapay zekâ ve görüntü işleme tekniklerinin hayvan türlerini tanımada güçlü bir araç olabileceğini; aynı zamanda biyolojik çeşitliliğin korunması, ekosistem sürdürülebilirliği ve bilimsel araştırmalar için hızlı ve maliyet etkin bir alternatif sunduğunu göstermektedir.

Keywords

Derin Öğrenme , Evcil hayvan görseli sınıflandırıcısı , Görüntü segmentasyonu , Topluluk öğrenme , Yapay zekâ

References

• Willi M, Pitman RT, Cardoso AW, Locke C, Swanson A, Boyer A, Fortson L (2019) Identifying animal species in camera trap images using deep learning and citizen science. Methods Ecol Evol 10:80–91. https://doi.org/10.1111/2041-210X.13099
• Norouzzadeh MS, Nguyen A, Kosmala M, Swanson A, Palmer MS, Packer C, Clune J (2018) Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning. Proc Natl Acad Sci USA 115:E5716–E5725. https://doi.org/10.1073/pnas.1719367115
• Ansari A, Ansari S, Prasla SS, Naveed A (2024) Comparative Analysis of Image Classification Methods on Cat Breeds and Behavior using Machine Learning Techniques. Pak J Eng Technol Sci 12(1):91–103. https://doi.org/10.22555/pjets.v12i1.1100
• Alfarhood S, Alrayeh A, Safran M, Alfarhood M, Che D (2023) Image-based Arabian camel breed classification using transfer learning on CNNs. Appl Sci 13(14):8192. https://doi.org/10.3390/app13148192
• Zhang S, Wang Y, Sun F, Yang Y (2024) Cat and dog breed classification based on SE-DenseNet integrated modeling. In: 2024 5th Int Conf on Information Science and Education (ICISE-IE). IEEE, pp 611–615.
• Tang J, Zhao Y, Feng L, Zhao W (2022) Contour-based wild animal instance segmentation using a few-shot detector. Animals 12:1980. https://doi.org/10.3390/ani12151980
• Ravi N, Gabeur V, Hu Y-T, Hu R, Ryali C, Ma T, Khedr H, Rädle R, Rolland C, Gustafson L, Mintun E, Pan J, Alwala KV, Carion N, Wu C-Y, Girshick R, Dollár P, Feichtenhofer C (2024) SAM 2: Segment Anything in Images and Videos. arXiv preprint arXiv:2408.00714.
• Kirillov A, Mintun E, Ravi N, Mao H, Rolland C, Gustafson L et al (2023) Segment anything. In: Proc IEEE/CVF Int Conf Comput Vis (ICCV), pp 4015–4026.
• Osco LP, Wu Q, De Lemos EL, Gonçalves WN, Ramos APM, Li J, Junior JM (2023) The segment anything model (SAM) for remote sensing applications: from zero to one shot. Int J Appl Earth Obs Geoinf 124:103540.
• Huang G, Liu Z, Van Der Maaten L, Weinberger KQ (2017) Densely connected convolutional networks. In: Proc IEEE Conf Comput Vis Pattern Recognit (CVPR), Honolulu, HI, USA, Jul 21–26, pp 4700–4708.
• Mobiny A, Singh A, Van Nguyen H (2019) Risk-aware machine learning classifier for skin lesion diagnosis. J Clin Med 8(8):1241.
• Nair K, Deshpande A, Guntuka R, Patil A (2022) Analysing X-ray images to detect lung diseases using DenseNet-169 technique. In: Proc 7th Int Conf Innovations and Research in Technology and Engineering (ICIRTE 2022), Mumbai, India, Apr 2022.
• Dalvi PP, Edla DR, Purushothama BR (2023) Diagnosis of coronavirus disease from chest X-ray images using DenseNet-169 architecture. SN Comput Sci 4:214. https://doi.org/10.1007/s42979-023-01662-7
• Szegedy C, Vanhoucke V, Ioffe S, Shlens J, Wojna Z (2016) Rethinking the inception architecture for computer vision. In: Proc IEEE Conf Comput Vis Pattern Recognit (CVPR), Las Vegas, NV, USA, Jun 27–30, pp 2818–2826.
• Li L, Chen Y, Shen Z, Zhang X, Sang J, Ding Y et al (2020) Convolutional neural network for the diagnosis of early gastric cancer based on magnifying narrow band imaging. Gastric Cancer 23(1):126–132.
• Ali L, Alnajjar F, Jassmi HA, Gocho M, Khan W, Serhani MA (2021) Performance evaluation of deep CNN-based crack detection and localization techniques for concrete structures. Sensors 21(5):1688.
• Xia X, Xu C, Nan B (2017) Inception-v3 for flower classification. In: 2nd Int Conf on Image, Vision and Computing (ICIVC), Chengdu, China, Jun 2–4, pp 783–787.
• Sam SM, Kamardin K, Sjarif NNA, Mohamed N (2019) Offline signature verification using deep learning convolutional neural network (CNN) architectures GoogLeNet inception-v1 and inception-v3. Procedia Comput Sci 161:475–483.
• Chollet F (2017) Xception: Deep learning with depthwise separable convolutions. In: Proc IEEE Conf Comput Vis Pattern Recognit (CVPR), Honolulu, HI, USA, Jul 21–26, pp 1251–1258.
• Mondal A, Samanta S, Jha V (2022) A convolutional neural network-based approach for automatic dog breed classification using modified-Xception model. In: Saeed K, Biswas A, Dai W et al (eds) Electronic Systems and Intelligent Computing. Springer, Singapore, pp 61–70.
• Bhoomika (2024) Precise image classification with Xception model. In: 2024 Second Int Conf on Intelligent Cyber Physical Systems and Internet of Things (ICoICI), Vellore, India, Aug 1–2, pp 1536–1540.
• Wu X, Liu R, Yang H, Chen Z (2020) An Xception-based convolutional neural network for scene image classification with transfer learning. In: 2020 2nd Int Conf on Information Technology and Computer Application (ITCA), Guangzhou, China, Dec 18–20, pp 262–267.
• Srinivasan K, Garg L, Datta D, Alaboudi AA, Jhanjhi NZ, Agarwal R, Thomas AG (2021) Performance comparison of deep CNN models for detecting driver’s distraction. Comput Mater Contin 68(3):4109–4124. https://doi.org/10.32604/cmc.2021.016736
• Köse B (2023) Veri, enformatik, yapay zeka ve optimizasyon. Kuantum Teknolojileri ve Enformatik Araştırmaları Dergisi 1:35–40.
• Polikar R (2012) Ensemble learning. In: Zhang C, Ma Y (eds) Ensemble Machine Learning: Methods and Applications. Springer, Boston, MA, pp 1–34.
• Rojarath A, Songpan W, Pong-inwong C (2016) Improved ensemble learning for classification techniques based on majority voting. In: 2016 7th IEEE Int Conf on Software Engineering and Service Science (ICSESS), Beijing, China, Aug 26–28, pp 107–110.
• Opitz D, Maclin R (1999) Popular ensemble methods: an empirical study. J Artif Intell Res 11:169–198.
• Dietterich TG (2000) Ensemble methods in machine learning. In: Kittler J, Roli F (eds) Multiple Classifier Systems. Springer, Berlin, Heidelberg, pp 1–15.
• Wu J, Chen XY, Zhang H, Xiong LD, Lei H, Deng SH (2019) Hyperparameter optimization for machine learning models based on Bayesian optimization. J Electron Sci Technol 17(1):26–40.
• Banda JM, Angryk RA, Martens PCH (2013) Steps toward a large-scale solar image data analysis to differentiate solar phenomena. Sol Phys 288(1):435–462.
• Hark C (2022) Sahte haber tespiti için derin bağlamsal kelime gömülmeleri ve sinirsel ağların performans değerlendirmesi. Fırat Üniv Müh Bilim Derg 34:733–742.
• Labani M, Moradi P, Ahmadizar F, Jalili M (2018) A novel multivariate filter method for feature selection in text classification problems. Eng Appl Artif Intell 70:25–37.
• Powers DM (2020) Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. arXiv preprint arXiv:2010.16061.