Wood defect detection and classification using deep learning / Yap Yi Ren
In the timber and wood industry, natural defects on wood and timber are always one of the main issues. In many timber and wood industry, the quality assurance of the board is still controlled by a human. This is because the defects can vary in many ways likes amount, shape, area and colour. The qual...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/11441/1/Yap_Yi_Ren.jpg http://studentsrepo.um.edu.my/11441/8/yi_ren.pdf http://studentsrepo.um.edu.my/11441/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In the timber and wood industry, natural defects on wood and timber are always one of the main issues. In many timber and wood industry, the quality assurance of the board is still controlled by a human. This is because the defects can vary in many ways likes amount, shape, area and colour. The quality checking process can be very tedious and worker may easily makes mistakes in judgement. To reduce the human mistakes, this study focuses on designing a wood defect detection and classification by using the artificial intelligence technique of Convolutional Neural Network (CNN) in MATLAB. Convolutional Neural Network (CNN) is one of the deep neural networks used in two-dimensional data. It mainly used to classify objects in images, cluster them by similarity and execute object recognition. This technology can identify faces, street sign, tumours, human, etc. The CNN model consists of input images, Convolution Layers, Activation Function (ReLU), Pooling, Fully Connected layers and Output layer. Three sets of input data such as Knots, Crack and Normal are prepared for training and testing the CNN model by using different parameters. The results of the different configurations are compared and analysed. The accuracy of overall classification is 97.2%. |
|---|