| 要約: | This research investigates the application ofConvolutional Neural Network (CNN) for palm oil Fresh FruitBunch (FFB) ripeness grading recognition. CNN has become thestate-of-the-art technique in computer vision especially in objectrecognition where the recognition accuracy is very impressive.Even though there is no need for feature extraction in CNN, itrequires a large amount of training data. To overcome thislimitation, utilising the pre-trained CNN model with transferlearning provides the solution. Thus, this research comparesCNN, pre-trained CNN model and hand-crafted feature andclassifier approach for palm oil Fresh Fruit Bunch (FFB)ripeness grading recognition. The hand-crafted features arecolour moments feature, Fast Retina Keypoint (FREAK) binaryfeature, and Histogram of Oriented Gradient (HOG) texturefeature with Support Vector Machine (SVM) classifier. Imagesof palm oil FFB with four different levels of ripeness have beenacquired, and the results indicate that with a small number ofsample data, pre-trained CNN model, AlexNet, outperformsCNN and the hand-crafted feature and classifier approach. © 2018 Universiti Teknikal Malaysia Melaka. All rights reserved.
|
|---|