An artificial neural network for real-time hardwood lumber grading

Abstract

a b s t r a c tComputerized grading of hardwood lumber according to NHLA rules would permit fast assessment of sawn lumber and the evaluation of potential edging and trimming operations to improve lumber value. More importantly, to enable optimization of the hardwood lumber sawing process, a fast means of evaluating the potential value of boards before they are sawn is necessary. As log and lumber scanning systems become prevalent and common, these needs become more pressing. From an automation perspective, the NHLA lumber grades are difficult to implement efficiently in a computer program. Exhaustive approaches that examine every potential cutting size and combination to determine the grade give accurate grading solutions, at the cost of computation time. Other approaches have examined heuristic methods that implement key parts of the grading rules, or used artificial neural network methods, both with the loss of accuracy. Here, a different approach to computerized grading is examined that takes a hybridized approach using projected yield from cut-up simulation and neural network methods. This new hybrid approach has the advantage of both accuracy and high-processing speed. Such an approach lends itself to log sawing optimization with respect to NHLA grades and market values when internal log defect information is known.Published by Elsevier B.V.

Automated Identification of Wood Veneer Surface Defects Using Faster Region-Based Convolutional Neural Network with Data Augmentation and Transfer Learning

“…The results with respect to different sliding window multipliers are presented in Table 2. The best accuracy, 76.6%, was achieved using the sliding window sizes of [4,8,12]. The classification accuracy using the pre-trained AlexNet, VGG16, BNInception, and ResNet152 neural network models trained with different network parameters (batch size and learning speed) is summarized in Figures 13-16.…”

“…The results with respect to different sliding window multipliers are presented in Table 2. The best accuracy, 76.6%, was achieved using the sliding window sizes of [4,8,12]. The best results while using the VGG16 neural network were achieved using 32 batch size and 0.01 learning speed for branch class, 32 batch size and 0.1 learning speed for scratch and background classes, 64 batch size and 0.2 learning speed for core class, and 100 batch size and 0.2 learning speed for background class.…”

“…Computer vision systems using high-resolution and high-speed video cameras scan products and report on product quality through various algorithms and techniques. Several methods can be used for surface analysis, such as Gabor [4] or wavelet [5] transforms, and deep learning techniques [6][7][8][9][10], but they are not always sufficiently accurate or fast. For example, the double threshold method is used to check the quality of materials in the textile industry [11] to reduce the amount of data in the images used for the training of neural networks.…”

Energy Prediction of Wheat Production Using Data Mining Technique in Iran

“…Some of these techniques, such as the kmeans, the k nearest neighbor, artificial neural networks and support vector machines, are discussed and an application in agriculture for each of these techniques is presented. In addition, the references included in brackets confirmed the ability and necessity of using data mining technique in agriculture (Raorane & Kulkarni, 2013;Kalpana et al, 2014;Geetha, 2015;Khedr et al, 2015;Raorane & Kulkarni, 2015;Almaliki et al, 2016;Almaliki, 2017;Oliveira et al, 2017;Thomas, 2017). Therefore, the energy used in wheat production was predicted and modelled using data mining techniques (Artificial Neural Network).…”

Derivation of NARX models by expanding activation functions in neural networks

“…Chen et al [3] and Levin and Narendra [4] demonstrated that nonlinear systems can be identified using neural networks. Furthermore, free open-source libraries such as the Fast Artificial Neural Network Library (FANN) [5] for network learning have already enabled researchers in various fields to use neural networks [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. In fact, neural networks have recently been used for the identification of a wide range of nonlinear systems, including biological systems [23][24][25][26][27][28][29][30][31][32][33][34][35][36].…”