KENALBATIK

Introduction

• The publication is titled under: Klasifikasi Motif Batik menggunakan metode Deep Convolutional Neural Network dengan Data Augmentation
• The web Application repo can be found here

Resnet-Batik-Classification

Classify Batik (Ceplok, Kawung, Lereng, Nitik, Parang, Lunglungan, and Semen) using Resnet50 and Resnet18.

also available VGG16 and VGG19 (weights=None) models to compare with Resnet.

What's inside:

• Preprocess Image.ipynb: to modify images with some methods (slicing image, augment image, or both).
• Create Dataset.ipynb: to Create Supervised Dataset which splitted into Train and Validation Dataset. Save in hdf5/h5py format.
• MAIN_SYSTEM: Folder contain all python files to training and evaluating model.

Folder MAIN_SYSTEM, How to Use main.py:

required arguments:

• model name options are: resnet50, resnet18, vgg16, or vgg19
• train file: h5py file contains training file
• validation file: h5py file contains validation file
• test file: h5py file contains test file
• number classes: number of labels/classes
• dropout: range 0 until 1 to dropout layer
• batch size recommended options: 8, 16, 32, 64
• lr_value recommended options: float type. between 1e-2 until 1e-6
• optimizer code options: 1 for Adam. 2 for SGD

how to begin training model:

python main.py [-h] [--test_file TEST_FILE] [--dropout, -d DROPOUT]
               [--epoch, -e EPOCH] [--class_number, -c CLASS_NUMBER]
               [--batch_size, -b BATCH_SIZE] [--optimizer, -o OPTIMIZER]
               [--lr_value, -lr LR_VALUE]
               model_name train_file val_file

examples:

python main.py resnet50 dataset/train.h5 dataset/val.h5

python main.py resnet50 dataset/train.h5 dataset/val.h5 -c5 -b32 -lr 1e-3

Requirement libraries

• Jupyter Notebook: recommended editor for .ipynb files
• Keras : for building Network and
• Imageio
• OpenCV2
• Numpy
• Matplotlib
• H5py
• sklearn

Poster