Recognition of Mushroom Species Using Few-Shot Learning Method with a Siamese Network

Year 2025, Volume: 9 Issue: 2, 262 - 266, 30.11.2025

Göktürk Öztürk , Köksal Erentürk

Abstract

Mushrooms, as nutritionally and medicinally valuable macrofungi, require accurate recognition due to the presence of toxic species causing severe health risks. Traditional methods based on morphology are time-consuming and prone to human error, which makes automated solutions essential. In this study, a siamese neural network with a ResNet18 backbone was applied to mushroom species recognition under a 7-way 3-shot learning setting. The dataset, derived from Kaggle, was pre-processed with background removal, resizing, normalization, and augmentation to ensure reliable feature extraction. The model was trained with cosine embedding loss and evaluated using accuracy, precision, recall, F1-score, and confusion matrix analysis. Results demonstrated a high classification accuracy of 90.48%, showing that the model effectively distinguishes mushroom species despite a small number of confusions. These findings confirm the effectiveness of siamese networks for mushroom classification and suggest future improvements.

Keywords

few-shot learning , siamese network , mushroom recognition , image classification , deep learning

Siamese Ağı ile Few-Shot Öğrenme Yöntemi Kullanılarak Mantar Türlerinin Tanınması

Abstract

Besinsel ve tıbbi açıdan değerli makrofunguslar olan mantarlar, ciddi sağlık risklerine neden olan zehirli türlerin varlığı nedeniyle doğru bir şekilde tanınmayı gerektirir. Morfolojiye dayalı geleneksel yöntemler zaman alıcıdır ve insan hatasına açıktır, bu da otomatik çözümleri gerekli kılar. Bu çalışmada, 7-way 3-shot öğrenme düzeninde, mantar türlerinin tanınması için ResNet18 omurgasına sahip bir siamese sinir ağı uygulanmıştır. Kaggle’dan elde edilen veri kümesi, güvenilir özellik çıkarımını sağlamak amacıyla arka plan kaldırma, yeniden boyutlandırma, normalleştirme ve veri artırma işlemleriyle ön işleme tabi tutulmuştur. Model, kosinüs gömme kaybı ile eğitilmiş ve doğruluk, kesinlik, duyarlılık, F1-skoru ve karmaşıklık matrisi analizleri kullanılarak değerlendirilmiştir. Sonuçlar, %90,48 gibi yüksek bir sınıflandırma doğruluğu ortaya koymuş ve modelin az sayıda karışıklık olmasına rağmen mantar türlerini etkili bir şekilde ayırt ettiğini göstermiştir. Bu bulgular, mantar sınıflandırması için siamese ağlarının etkinliğini doğrulamakta ve gelecekteki iyileştirmelere işaret etmektedir.

Keywords

few-shot öğrenme , siamese ağı , mantar tanıma , görüntü sınıflandırma , derin öğrenme

References

• [1] F. Kalyoncu and M. Oskay, “Antimicrobial activities of four wild mushroom species collected from Turkey,” in Proceedings of the 6th International Conference on Mushroom Biology and Mushroom Products, Bonn, Germany, 2008, pp. 31–35.
• [2] D. L. Hawksworth, “Mushrooms: the extent of the unexplored potential,” International Journal of Medicinal Mushrooms, vol. 3, no. 4, 2001.
• [3] W. Yun and I. R. Hall, “Edible ectomycorrhizal mushrooms: challenges and achievements,” Canadian Journal of Botany, vol. 82, no. 8, pp. 1063–1073, 2004.
• [4] M. Janatolmakan, M. R. Ganji, T. Ahmadi-Jouybari, S. Rezaeian, M. Ghowsi, and A. Khatony, “Demographic, clinical, and laboratory findings of mushroom-poisoned patients in Kermanshah province, west of Iran,” BMC Pharmacology and Toxicology, vol. 23, no. 1, pp. 72, 2022.
• [5] R. N. Bashir, O. Mzoughi, N. Riaz, M. Mujahid, M. Faheem, M. Tausif, and A. R. Khan, “Mushroom species classification in natural habitats using convolutional neural networks (CNN),” IEEE Access, vol. 12, pp. 176818-176832, 2024.
• [6] A. Gobinath, P. Rajeswari, and M. Anandan, “Automated mushroom species identification and prediction using image processing and machine learning,” in Proc. 2024 Int. BIT Conf. (BITCON), Dhanbad, India, 2024, pp. 1–5.
• [7] A. Wibowo, Y. Rahayu, A. Riyanto, and T. Hidayatulloh, “Classification algorithm for edible mushroom identification,” in Proc. International Conference on Information and Communications Technology (ICOIACT), Yogyakarta, Indonesia, 2018, pp. 250–253.
• [8] M. H. Ahmad, S. R. Sohag, S. Bormon, H. R. Shishir, and S. H. Antar, “AI-driven mycology: transfer learning for automated mushroom species identification,” in Proc. IEEE Int. Conf. Computing, Applications and Systems (COMPAS), Cox’s Bazar, Bangladesh, Sep. 2024, pp. 1–9.
• [9] A. H. Rangkuti, J. M. Kerta, B. Juarto, A. Kurniawan, A. Tandianto, and M. Gabriela, “Refining mushroom image recognition using deep learning and image similarity distance,” in Proc. 2024 Int. Conf. Informatics, Multimedia, Cyber and Inf. Syst. (ICIMCIS), Jakarta, Indonesia, Nov. 2024, pp. 531–536.
• [10] E. Özbay, F. A. Özbay, and F. S. Gharehchopogh, “Visualization and classification of mushroom species with multi-feature fusion of metaheuristics-based convolutional neural network model,” Applied Soft Computing, vol. 164, Art. no. 111936, 2024.
• [11] Y. Peng, Y. Xu, J. Shi, and S. Jiang, “Wild mushroom classification based on improved mobilevit deep learning,” Applied Sciences., vol. 13, no. 8, Art. no. 4680, 2023.
• [12] S. Subramani, A. F. Imran, T. T. M. Abhishek, K. M. Sanjay, and J. Yaswanth, “Deep learning based detection of toxic mushrooms in Karnataka,” Procedia Computer Science, vol. 235, pp. 91-101, 2024.
• [13] Z. Wei, J. Wang, H. You, R. Ji, F. Wang, L. Shi, and H. Yu, “A lightweight context-aware framework for toxic mushroom detection in complex ecological environments,” Ecological Informatics, vol. 90, Art. no. 103256, 2025.
• [14] M. Du, F. Wang, W. Yan, J. Guo, L. Liu, P. Lv, Y. He, X. Feng, and Y. Wang, “Improving food safety: Synthetic data augmentation for accurate mushroom species identification in complex environments,” Applied Food Research, vol. 5, no. 1, Art. no. 101039, 2025.
• [15] L. Fei-Fei, “A Bayesian approach to unsupervised one-shot learning of object categories,” in Proc. 9th IEEE International Conference on Computer Vision (ICCV), Nice, France, Oct. 2003, pp. 1134–1141.
• [16] L. Fei-Fei, R. Fergus, and P. Perona, “One-shot learning of object categories,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, no. 4, pp. 594–611, Apr. 2006.
• [17] G. Koch, R. Zemel, and R. Salakhutdinov, “Siamese neural networks for one-shot image recognition,” in Proc. ICML Deep Learning Workshop, vol. 2, no. 1, pp. 1–30, 2015.
• [18] Z. Guo, Y. Wang, L. Liu, S. Sun, B. Feng, and X. Zhao, “Siamese network-based few-shot learning for classification of human peripheral blood leukocyte,” in Proc. 2021 IEEE 4th International Conference on Electronic Information and Communication Technology (ICEICT), Xi'an, China, Aug. 2021, pp. 818–822.
• [19] I. A. Lungu, A. Aimar, Y. Hu, T. Delbruck, and S. -C. Liu, “Siamese networks for few-shot learning on edge embedded devices,” IEEE Journal on Emerging Selected Topics in Circuits and Systems, vol. 10, no. 4, pp. 488–497, 2020.
• [20] Kaggle Contributors, “Mushrooms images classification 215,” Kaggle, [Online]. Available: https://www.kaggle.com/datasets/daniilonishchenko/mushrooms-images-classification-215 [Accessed: Sep. 8, 2025].
• [21] Pixelcut, “Background Remover Tool,” [Online]. Available: https://www.pixelcut.ai/background-remover [Accessed: Sep. 8, 2025].