Text
IMPLEMENTASI CONVOLUTIONAL NEURAL NETWORK (CNN) UNTUK KLASIFIKASI CITRA NATURAL SCENE
Humans recognized natural scenes using their sight. Natural scene problems appear when applied to navigation robots, map recognition, and automatic surveillance systems. Researchers developed a software that can classify natural scene images using Convolutional Neural Network (CNN). The CNN method used in this study compares three architectures, namely ResNet50V2, VGG16, and EfficientNetB4. The models were trained with an image dataset which divided into 10902 training data, 2725 validation data and 3407 test data. There are six combinations of learning rate and batch size for tuning the best model, namely learning rate 0.001 batch size 12, learning rate 0.01 batch size 12, learning rate 0.01 batch size 10, learning rate 0.001 batch size 10, learning rate 0.01 batch size 8, and learning rate 0.01 batch size 8. The test results show that best architecture for natural scene image classification is EfficientNetB4 which obtains an accuracy value of 93% with learning rate 0.001 and batch size 8
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2207005282 | T84883 | T848832022 | Central Library (Referens) | Available but not for loan - Not for Loan |
No other version available