Machine Learning Jual.In

Machine learning model for image classification and location-based recommendation system in Jual.In application

Contributing

Team Machine Learning on Jual.in:

• Satriadi Putra Santika
• Fhillipus Mahendra

Dataset

We import the dataset from kaggle.com and web scrapping in dataumkm.com and Google Image

• Fashion Dataset

Library

Library we use in this model:

  tensorflow
  os
  numpy
  pandas
  cv2
  splitfolders
  scikit-learn
  matplotlib
  seaborn

Image Classification

This classification has 25 classes which is:

• 0: 'Ayam Goreng Tepung'
• 1: 'Baju'
• 2: 'Baso'
• 3: 'Celana'
• 4: 'Dompet'
• 5: 'Gaun'
• 6: 'Jam Tangan'
• 7: 'Jus'
• 8: 'Kacamata'
• 9: 'Kaos Kaki'
• 10: 'Keripik'
• 11: 'Kopi'
• 12: 'Makeup'
• 13: 'Martabak Manis'
• 14: 'Mie Ayam'
• 15: 'Mie Goreng'
• 16: 'Nasi Goreng'
• 17: 'Parfum'
• 18: 'Pisang Goreng'
• 19: 'Roti Bakar'
• 20: 'Sate Ayam'
• 21: 'Sendal'
• 22: 'Sepatu'
• 23: 'Tas'
• 24: 'Topi'

We use transfer learning method Xception on our model so it can reach high accuracy.

Accuracy before Fine-Tuning:

Accuracy after Fine-Tuning:

Result predict on image: