AlviGeo/batik-motif-resnet50

BatikAI — Indonesian Batik Motif Classifier

Fine-tuned ResNet-50 on the Batik Indonesia dataset.

Labels (10 motifs)

parang, kawung, mega_mendung, truntum, sekar_jagad, ceplok, sidomukti, lereng, nitik, sogan

Usage

from inference import load_model, predict
from PIL import Image

model, device, transform, class_names, config = load_model()
img = Image.open("batik.jpg")
predictions, info = predict(img, model, device, transform, class_names)
print(predictions)

Training

• Architecture: ResNet-50 (ImageNet pretrained, fine-tuned)
• Optimizer: AdamW (lr=1e-4, weight_decay=1e-3)
• Scheduler: CosineAnnealingLR (15 epochs)
• Input size: 224×224
• Augmentation: RandomFlip, Rotation, ColorJitter, RandomPerspective

Live Demo

🚀 Try the Gradio Demo