THE FRACTAL NEURAL NETWORK APPROACH FOR WOOD SPECIES CLASSIFICATION
DOI:
https://doi.org/10.15588/1607-3274-2026-2-11Keywords:
wood species recognition, fractal neural network, self similarity, data augmentation, UAVAbstract
Context. Automated wood species recognition from macroscopic sections must combine high accuracy with stringent computational and memory requirements typical of peripheral devices and UAV onboard platforms. The heterogeneous, partially self-similar structure of wood texture, combining self-similar mid-level fluctuations with locally high-contrast elements and a large spread of leaders reflecting multifractality, motivates the creation of a fractal neural network architecture that can encode multiscale patterns while remaining computationally efficient.
Objective. The goal is to develop and validate fractal neural network models that achieve high recognition quality at reduced computational cost, enabling practical deployment on systems with limited resources.
Method. A fractal neural architecture (FractalNet) is employed to realize a self similar multi branch topology that forms an ensemble of receptive fields from fine to coarse scales, coherently modeling the spectrum of local regularities captured by leaders. The approach is benchmarked against ResNet50 and VGG16 with and without data augmentation on a 12 class macroscopic image task. Evaluation includes per class precision, recall, and F1, macro/weighted aggregates, confusion matrices, and analysis of numerical complexity in terms of trainable parameters and layer depth to assess deployability.
Results. The FractalNet combined with augmentation procedure attains the best overall performance, reaching macro F1 = 0.80, weighted F1 = 0.81, and accuracy = 0.81, outperforming ResNet50 (macro F1 = 0.57) and VGG16 (macro F1 = 0.71). Confusion matrices exhibit reduced cross class confusions, indicating more uniform gains across species. Despite superior accuracy, FractalNet contains ~0.37 M parameters versus 23.6 M in ResNet50 and 65.1 M in VGG16, yielding a markedly smaller memory footprint and lower inference latency.
Conclusions. The alignment between multifractal texture properties of wood and the self similar design of FractalNet produces a favorable accuracy-efficiency trade off. The method delivers state of the art recognition quality while preserving computational frugality, thus enabling reliable use in resource constrained scenarios, including UAV based and other edge deployments
References
Yang X., Zheng Z., Zheng H., Liu X. Deep learning method of precious wood image classification based on microscopic computed tomography. Russ J Nondestruct Test, 2024, Vol. 60, № 10, pp. 1136–1148. DOI: 10.1134/S1061830924602447
He T., Mu S., Zhou H., Hu J. Wood species identification based on an ensemble of deep convolution neural networks. WR, 2021, Vol. 66, № 1, pp. 1–14. DOI: 10.37763/wr.1336- 4561/66.1.0114
Wang C.-K., Zhao P. Classification of wood species using spectral and texture features of transverse section. Eur. J. Wood Prod, 2021, Vol. 79, № 5, pp. 1283–1296. DOI: 10.1007/s00107-021-01728-9
Berezsky O., Liashchynskyi P., Pitsun O., Izonin I. Synthesis of Convolutional Neural Network architectures for biomedical image classification. Biomedical Signal Processing and Control, 2024, Vol. 95, pp. 106325. DOI: 10.1016/j.bspc.2024.106325
Berezsky O. M., Liashchynskyi P. B., Pitsun O. Y., Melnyk G. M. Deep network-based method and software for small sample biomedical image generation and classification. RIC, 2024, № 4, P. 76. DOI: 10.15588/1607-3274- 2023-4-8
Chapaliuk B. V., Zaychenko Y. P. Medical image segmentation methods overview. SRIT, 2018, Vol. 0, № 1, P. 72–81. DOI: 10.20535/SRIT.2308-8893.2018.1.05
Kim J. Deep learning-based vehicle type and color classification to support safe autonomous driving. Applied Sciences, 2024, Vol. 14, № 4, P. 1600. DOI: 10.3390/app14041600
Hu Z., Bodyanskiy Y. V., Kulishova N. Ye., Tyshchenko O. K. A multidimensional extended neo-fuzzy neuron for facial expression recognition. IJISA, 2017, Vol. 9, № 9, P. 29–36. DOI: 10.5815/ijisa.2017.09.04
Nemavhola A., Chibaya C., Viriri S. A systematic review of CNN architectures, databases, performance metrics, and applications in face recognition. Information, 2025, Vol. 16, № 2, P. 107. DOI: 10.3390/info16020107
Ke H. Using convolutional neural networks for material surface quality inspection and classification. Int J Comput Intell Syst, 2025, Vol. 18, № 1, P. 224. DOI: 10.1007/s44196-025-00951-z
Duvvuri K., Kanisettypalli H., Jaswanth K., Murali K.Video classification using CNN and ensemble learning. 9th International Conference on Advanced Computing and Communication Systems (ICACCS-2023), Coimbatore : proceedings, Coimbatore : IEEE, 2023, pp. 66–70. DOI: 10.1109/ICACCS57279.2023.10112975
Zheng C. et al.Opening the black box: explainable deeplearning classification of wood microscopic image of endangered tree species. Plant Methods, 2024, Vol. 20, № 1, P. 56. DOI: 10.1186/s13007-024-01191-6
Liu S. et al.Automated species discrimination and feature visualization of closely related Pterocarpus wood species using deep learning models: comparison of four convolutional neural networks. Wood Sci Technol, 2025, Vol. 59, № 5, P. 86. DOI: 10.1007/s00226-025-01690-2
Kim T. K. et al. Identifying and extracting bark key features of 42 tree species using convolutional neural networks and class activation mapping. Sci Rep, 2022, Vol. 12, № 1, P. 4772. DOI: 10.1038/s41598-022-08571-9
Kanayama H., Ma T., Tsuchikawa S., Inagaki T. Cognitive spectroscopy for wood species identification: near infrared hyperspectral imaging combined with convolutional neural networks. Analyst, 2019, Vol. 144, № 21, pp. 6438–6446. DOI: 10.1039/C9AN01180C
Pan X., Yu Z., Yang Z. A multi-scale convolutional neural network combined with a portable near-infrared spectrometer for the rapid, non-destructive identification of wood species. Forests, 2024, Vol. 15, № 3, P. 556. DOI: 10.3390/f15030556
Park S.-Y. Kim J.-H., Kim J.-C., Yang S.-Y., Yeo H., Choi I.-G. Classification of softwoods using wood extract information and near infrared spectroscopy. BioRes, 2021, Vol. 16, № 3, pp. 5301–5312. DOI: 10.15376/biores.16.3.5301-5312
Novaes T. V. et al. Discrimination of Amazonian forest species by NIR spectroscopy: wood surface effects. Eur. J. Wood Prod, 2023, Vol. 81, № 1, P. 159–172. DOI: 10.1007/s00107-022-01862-y
He Y., Wang Y., Ma W. A multiple spectral important feature fusion method for wood species identification. Wood Sci Technol, 2025, Vol. 59, № 2, P. 34. DOI: 10.1007/s00226-025-01639-5
Li C., Wang Y. Optimizing recognition models for wood species identification using multi-spectral techniques. Holzforschung, 2025, Vol. 79, № 4–5, pp. 177–187. DOI: 10.1515/hf-2024-0112
Li S., Li D., Yuan W.Wood defect classification based on two-dimensional histogram constituted by LBP and local binary differential excitation pattern. IEEE Access, 2019, Vol. 7, pp. 145829–145842. DOI: 10.1109/ACCESS.2019.2945355
Rahiddin R. N. N., Hashim U. R., Ismail N. H., Salahuddin L., Choon N. H., Zabri S. N. Classification of wood defect images using local binary pattern variants. Int. J. Adv. Intell. Informatics, 2020, Vol. 6, № 1, P. 36. DOI: 10.26555/ijain.v6i1.392
Rahiddin R. N. N., Hashim U. R., Salahuddin L., Kanchymalay K., Wibawa A. P., Chun T. H. Local texture representation for timber defect recognition based on variation of LBP, IJACSA, 2022, Vol. 13, № 10. DOI: 10.14569/IJACSA.2022.0131053
Poelzleitner W., Schwingskakl G. Quality classification of wooden surfaces using Gabor filters and genetic feature optimization. Photonics East ’99 : proceedings, Ed.: D. P. Casasent. Boston, 1999, pp. 220–231. DOI: 10.1117/12.360301
Yeh M.-F., Luo C.-C., Liu Y.-C. Optimization of Gabor convolutional networks using the Taguchi method and their application in wood defect detection. Applied Sciences, 2025, Vol. 15, № 17, P. 9557. DOI: 10.3390/app15179557
Porter R., Canagarajah N. Robust rotation-invariant texture classification: wavelet, Gabor filter and GMRF based schemes. IEE Proc., Vis. Image Process, 1997, Vol. 144, № 3, P. 180. DOI: 10.1049/ip-vis:19971182
Florindo J. B., Bruno O. M. Fractal descriptors based on Fourier spectrum applied to texture analysis, 2012. DOI: 10.48550/arXiv.1201.4597
Wang F., Li Z.-S., Liao G.-P.Multifractal detrended fluctuation analysis for image texture feature representation. Int. J. Patt. Recogn. Artif. Intell, 2014, Vol. 28, № 03, P. 1455005. DOI: 10.1142/S0218001414550052
Tibebu D. T., Avramidis S. Fractal dimension of wood pores from pore size distribution. Holzforschung, 2022, Vol. 76, № 11–12, P. 967–976. DOI: 10.1515/hf-2021-0175
Subbotin S. Methods of data sample metrics evaluation based on fractal dimension for computational intelligence model building. 4th International Scientific-Practical Conference Problems of Infocommunications. Science and Technology (PIC S&T), Kharkov : proceedings. Kharkov, IEEE, 2017, pp. 1–6. DOI: 10.1109/INFOCOMMST.2017.8246136
He X. et al. Machine learning-based wood anatomy identification: towards anatomical feature recognition. IAWA J, 2024, Vol. 45, № 4, pp. 457–475. DOI: 10.1163/22941932- bja1015732. Larsson G., Maire M., Shakhnarovich G. FractalNet: Ultradeep neural networks without residuals. arXiv, 2016. DOI: 10.48550/ARXIV.1605.07648
Shymanskyi V., Ratinskiy O., Shakhovska N. Fractal neural network approach for analyzing satellite images. Applied Artificial Intelligence, 2025, Vol. 39, № 1, P. 2440839. DOI: 10.1080/08839514.2024.2440839
Shakhovska N., Shymanskyi V., Prymachenko M.FractalNet-LSTM model for time series forecasting. CMC, 2025, Vol. 82, № 3, pp. 4469–4484. DOI: 10.32604/cmc.2025.062675
Lin D. C., Sharif A. Wavelet transform modulus maxima based fractal correlation analysis, 2007. DOI: 10.1140/epjb/e2008-00004-6
Jaffard S., Lashermes B., Abry P. Wavelet leaders in multifractal analysis. Wavelet Analysis and Applications : proceedings, Eds.: T. Qian, M. I. Vai, Y. Xu. Basel, Birkhäuser Basel, 2007, pp. 201–246.
Barmpoutis P. Wood species dataset [Electronic resource]. Zenodo, 2019. DOI: 10.5281/ZENODO.2545611
Barmpoutis P., Dimitropoulos K., Barboutis I., Grammalidis N., Lefakis P. Wood species recognition through multidimensional texture analysis. Computers and Electronics in Agriculture, 2018, Vol. 144, pp. 241–248. DOI: 10.1016/j.compag.2017.12.011
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 V. M. Shymanskyi, N. B. Shakhovska
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Creative Commons Licensing Notifications in the Copyright Notices
The journal allows the authors to hold the copyright without restrictions and to retain publishing rights without restrictions.
The journal allows readers to read, download, copy, distribute, print, search, or link to the full texts of its articles.
The journal allows to reuse and remixing of its content, in accordance with a Creative Commons license СС BY -SA.
Authors who publish with this journal agree to the following terms:
-
Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License CC BY-SA that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
-
Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
-
Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) as it can lead to productive exchanges, as well as earlier and greater citation of published work.
