An Efficient Multiple Convolutional Neural Network Model (MCNN-14) for Fashion Image Classification

Authors: Sajjad Saed, Babak Teimourpour*, Kamand Kalashi, Mohammad Ali Soltanshahi
Affiliation: Department of Information Technology Engineering, School of Systems and Industrial Engineering, Tarbiat Modares University (TMU), Tehran, Iran

This repository contains the implementation of MCNN-14, a Multiple Convolutional Neural Network model proposed in our paper:

Saed et al. “An Efficient Multiple Convolutional Neural Network Model (MCNN-14) for Fashion Image Classification.”. ICWR2024

---

📌 Abstract

The transformation of fashion through online platforms has spurred a need for high-quality clothing search engines, facilitating seamless product discovery for global consumers. However, this transition has brought forth challenges in categorization and description standards among retailers and search engines, stemming from the inherent complexity and variability of fashion items. To address these challenges, deep learning techniques like Multiple Convolutional Neural Networks (MCNNs) have gained prominence in the fashion industry. We propose MCNN-14, a novel multiple-CNN architecture that balances superior classification accuracy with computational efficiency.

✨ Key Contribution: Our model achieved 93.08% accuracy on the Fashion-MNIST dataset, surpassing existing benchmarks.

---

👩‍💻 Authors

Sajjad Saed

Dr.Babak Teimourpour

Kamand Kalashi

Dr.Mohamad Ali Soltanshahi

---

📊 Performance

Model	Accuracy (%)
CNNs [32]	92.87
LSTM [33]	89.00
LeNet [34]	90.16
LSTMs [35]	88.26
VGG [36]	92.30
CNN LeNet-5 [37]	90.64
SVM+HOG [38]	88.53
ViT [39]	90.98
MCNN-14 (Ours)	93.08 (SOTA)

---

🚀Features

• Multiple CNN (MCNN-14) architecture
• Optimized for Fashion-MNIST
• Achieved 93.08% classification accuracy
• Balances accuracy with computational efficiency
• Built with TensorFlow / Keras

---

📂 Datasets

This project uses the Fashion-MNIST dataset, a widely-used benchmark dataset for clothing image classification.

• Description: Fashion-MNIST consists of 70,000 grayscale images of fashion items across 10 categories, with 28x28 pixel resolution.
• Train/Test Split: 60,000 training images and 10,000 test images.
• Source: Directly available via Keras datasets, automatically downloaded when using:

from tensorflow.keras.datasets import fashion_mnist
(x_train, y_train), (x_test, y_test) = fashion_mnist.load_data()

---

📚Citation

If you use the datasets or findings from our paper, please cite our paper in your work:

@INPROCEEDINGS{10533341,
  author={Saed, Sajjad and Teimourpour, Babak and Kalashi, Kamand and Soltanshahi, Mohammad Ali},
  booktitle={2024 10th International Conference on Web Research (ICWR)}, 
  title={An Efficient Multiple Convolutional Neural Network Model (MCNN-14) for Fashion Image Classification}, 
  year={2024},
  volume={},
  number={},
  pages={13-21},
  keywords={Computational modeling;Clothing;Computer architecture;Search engines;Benchmark testing;Feature extraction;Computational efficiency;Deep Learning;Image Classification;Multiple Convolutional Neural Networks;Fashion-MNIST},
  doi={10.1109/ICWR61162.2024.10533341}}

---

📝 License

This project is licensed under the MIT License. See the LICENSE file for more details.

If you use this repository, please mention the original GitHub repository by linking to MCNN-14-Fashion-Image-Classification. This helps support the project and acknowledges the contributors.