BACNN: Multi-scale feature fusion-based bilinear attention convolutional neural network for wood NIR classification

Abstract

Effective development and utilization of wood resources is critical. Wood modification research has become an integral dimension of wood science research, however, the similarities between modified wood and original wood render it challenging for accurate identification and classification using conventional image classification techniques. So, the development of efficient and accurate wood classification techniques is inevitable. This paper presents a one-dimensional, convolutional neural network (i.e., BACNN) that combines near-infrared spectroscopy and deep learning techniques to classify poplar, tung, and balsa woods, and PVA, nano-silica-sol and PVA-nano silica sol modified woods of poplar. The results show that BACNN achieves an accuracy of 99.3% on the test set, higher than the 52.9% of the BP neural network and 98.7% of Support Vector Machine compared with traditional machine learning methods and deep learning based methods; it is also higher than the 97.6% of LeNet, 98.7% of AlexNet and 99.1% of VGGNet-11. Therefore, the classification method proposed offers potential applications in wood classification, especially with homogeneous modified wood, and it also provides a basis for subsequent wood properties studies.

References

[] AcquarelliJ, van LaarhovenT, GerretzenJ, TranTN, BuydensMarchioriLME. Convolutional neural networks for vibrational spectroscopic data analysis. Anal Chim Acta, 2017, 954: 22-31

[] Asadi K, Littman M L (2017) In: An alternative softmax operator for reinforcement learning. In: Proc. 28th int’l conf. mach. Learn. Bellevuepp, WA, 243–252

[] ChenYY, WangZB. Quantitative analysis modeling of infrared spectroscopy based on ensemble convolutional neural networks. Chemometr Intell Lab Syst, 2018, 181: 1-10

[] GaoSH, HanQ, LiD, ChenMM, PengP. Representative batch normalization with feature calibration. Virtual, 2021, 1: 8669-8679

[] GaoMY, WangF, SongP, LiuJY, QiDW. BLNN: multiscale feature fusion-based bilinear fine-grained convolutional neural network for image classification of wood knot defects. J Sens, 2021, 2021: 1-18

[] Graham B, Engelhardt B, Van den Oord A (2014) Fractional max-pooling. arXiv preprint arXiv:1412.6071

[] HeKM, ZhangXY, RenSQ, SunJ. Deep residual learning for image recognition. CVPR, 2016 Las Vegas Nevada 770-778

[] HuC, QuJJ, XuCP, ZhuAJ. Garment image recognition based on adaptive pooling neural network. J Comput Appl, 2018, 38(8): 2211

[] HuJ, ShenL, SunG. Squeeze-and-excitation networks, 2018 Salt Lake City Utah CVPR 7132-7141

[] HuangPG, FanZ, LiXP, GuanC, ZhangYF, WuZK. Review of computer-based wood feature extraction and identification. World for Res, 2020, 33(01): 44-48

[] Huang G, Liu Z, Van Der Maaten L, Weinberger Kilian Q (2017) Densely Connected Convolutional Networks. CVPR 2017. Honolulu Hawaii. 4700–4708

[] JiaWS, ZhangHZ, MaJ, LiangG, WangJH, LiuX. Study on the predication modeling of COD for water based on UV-VIS spectroscopy and CNN algorithm of deep learning. Spectrosc Spectr Anal, 2020, 40(9): 2981

[] JiaoLC, ZhangF, LiuF, YangSY, LiLL, FengZX, QuR. A survey of deep learning-based object detection. IEEE Access, 2019, 7: 128837-128868

[] KauppinenJK. Fourier Self-Deconvolution in Spectroscopy. Spectrom Tech, 1983, 1983: 199-232

[] KrizhevskyA, SutskeverI, HintonG. Imagenet classification with deep convolutional neural networks. Commun ACM, 2012, 60(6): 84-90

[] KueselAC, StoyanovaR, AikenNR, LiCW, SzwergoldBS, ShallerC, BrownTR. Quantitation of resonances in biological 31P NMR spectra via principal component analysis: potential and limitations. NMR Biomed, 1996, 9(3): 93-104

[] LaakmannF, PetersenP. Efficient approximation of solutions of parametric linear transport equations by ReLU DNNs. Adv Comput Math Dio, 2021

[] LecunY, BottouL, BengioY. Gradient-based learning applied to document recognition. Proc IEEE, 1998, 86(11): 2278-2324

[] LecunY, BengioY, HintonG. Deep learning. Nature, 2015, 521(7553): 436-444

[] MaHY, LiXH, QiangL, XieXB, ChongJ. Zhao CX (2020) Research on identification technology of explosive vibration based on EEMD energy entropy and multiclassification SVM. Shock Vib, 2020, 2: 1-10

[] MaciorA, ZaborniakI, ChmielarzP, SmendaJ, WolskiK, CiszkowiczE, Lecka-SzlachtaK. A new protocol for ash wood modification: synthesis of hydrophobic and antibacterial brushes from the wood surface. Molecules, 2022, 27(3): 890

[] NisgoskiS, DeOliveiraA, DeMuñizG. Artificial neural network and SIMCA classification in some wood discrimination based on near-infrared spectra. Wood Sci Technol, 2017, 51(4): 929-942

[] PachutaSJ. Enhancing and automating TOF-SIMS data interpretation using principal component analysis. Appl Surf Sci, 2004, 231(6): 217-223

[] PanX, QiuJ, YangZ. Identification of softwood species using convolutional neural networks and raw near-infrared spectroscopy. Wood Mater Sci Eng, 2022, 1: 1-11

[] PradhanT, KumarP, PalS. CLAVER: an integrated framework of convolutional layer, bidirectional LSTM with attention mechanism based scholarly venue recommendation. Inf Sci, 2021, 559: 212-235

[] QinYH, DingXQ, GongHL. High dimensional feature selection in near infrared spectroscopy classification. Infrared Laser Eng, 2013, 42(5): 1355-1359

[] RumelhartDE, HintonGE, WilliamsRJ. Learning representations by back-propagating errors. Nature, 1986, 323(6088): 533-536

[] Simonyan K, Zisserman A (2014) Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv, https://arxiv.org/abs/1409.1556.

[] SoaresSFC, MedeirosEP, PasquiniC. Classification of individual cotton seeds with respect to variety using near-infrared hyperspectral imaging. Anal Methods, 2016, 8(48): 8498-8505

[] SzegedyC, LiuW, JiaYQ, SermanetP, ReedS, AnguelovD, RabinovichA. Going deeper with convolutions, 2015 Boston Massachusetts CVPR 1-9

[] TangYS, ChenZG. Soil pH prediction based on convolution neural network and near infrared spectroscopy. Spectrosc Spectr Anal, 2021, 41(3): 892

[] WangXS, SunYD, HuangAM. Research on infrared spectrum for timber species identification. For Eng, 2015, 31(6): 65-70

[] WangXS, SunYD, HuangMG, HuangAM. Back propagation artificial neural network combined with near infrared spectroscopy for timber recognition. J Northeast for Univ, 2015, 43(12): 82-85

[] WangQQ, GaoQX, GaoXB, NieFP. l(2, p)-Norm based PCA for image recognition. IEEE Trans Image Process, 2017, 27(3): 1336-1346

[] WangWQ, ZhangJ, WangFL. Attention bilinear pooling for fine-grained classification. Symmetry, 2019, 11(8): 1033

[] XiaJJ, HuangY, LiQQ, XiongYM, MinSG. Convolutional neural network with near-infrared spectroscopy for plastic discrimination. ECL, 2021, 19(5): 3547-3555

[] YangSY, KwonO, ParkY, ChungH, KimH, ParkSY, ChoiIG, YeoH. Application of neural networks for classifying softwood species using near infrared spectroscopy. J near Infrared Spectrosc, 2020, 28(5–6): 298-307

[] ZhangW, LiCH, PengGL, ChenYH, ZhangZJ. A deep convolutional neural network with new training methods for bearing fault diagnosis under noisy environment and different working load. Mech Syst Signal Process, 2017, 100: 439-453