A5027147345- Industrial Vision Systems and Defect Detection
- Remote Sensing and LiDAR Applications
- Wood and Agarwood Research
- Infrastructure Maintenance and Monitoring
- Surface Roughness and Optical Measurements
- Wood Treatment and Properties
Technical University of Malaysia Malacca
2022
This paper discusses the efficacy of data augmentation method deployed in many Convolutional Neural Network (CNN) algorithms for determining timber defect four species from Malaysia. A sequence morphological transformation, involving x-reflection and rotation, was executed dataset aiding CNN model training generating finest models which offer best classification performance defect. For further assessing algorithms' performance, several deep learning hyperparameters were tried on Merbau by...
This study proposed a classification model for timber defect based on an artificial neural network (ANN). Besides that, the research also focuses determining appropriate parameters in optimizing identification performance, such as number of hidden layers nodes and epochs network. The network's performance is compared with other standard classifiers Naïve Bayes, K-Nearest Neighbours, J48 Decision Tree finding their significant differences across multiple species. classifier's measured...
This paper evaluates timber defect classification performance across four various Local Binary Patterns (LBP). The light and heavy used in the study are Rubberwood, KSK, Merbau, Meranti, eight natural defects involved; bark pocket, blue stain, borer holes, brown knot, rot, split, wane. A series of LBP feature sets were created by employing Basic LBP, Rotation Invariant Uniform a phase extraction procedures. Several common classifiers to further separate classes, which Artificial Neural...
This study investigates the potential enhancement of classification accuracy in timber defect identification through utilization deep learning, specifically residual networks. By exploring refinement these networks via increased depth and multi-level feature incorporation, goal is to develop a framework capable distinguishing various classes. A sequence ablation experiments was conducted, comparing our proposed framework's performance (R1, R2 R3) with original ResNet50 architecture....