FashionVision: Intel-Optimized Clothing Recognition
- 0 Collaborators
FashionVision: Revolutionizing Clothing Recognition with Intel's OpenVINO Optimization. Our CNN model achieves 94.5% accuracy on Fashion MNIST, while OpenVINO cuts inference time by 97%. Explore our streamlined fashion recognition solution now! ...learn more
Project status: Published/In Market
Artificial Intelligence, Intel® Unnati
Intel Technologies
DevCloud,
oneAPI,
Intel Python
Overview / Usage
The project, "FashionVision," addresses the challenge of accurate clothing recognition with real-world applications in e-commerce, fashion design, and retail industries. It leverages a Convolutional Neural Network (CNN) model and Intel's OpenVINO optimization to achieve remarkable accuracy and significantly reduce inference time. This research empowers automated fashion recommendation systems, efficient inventory management, and enhanced image-based search engines.
Methodology / Approach
Our approach combines cutting-edge technology with efficient problem-solving techniques. We employ TensorFlow and Keras for developing the CNN model. The model architecture includes convolutional layers, max-pooling, batch normalization, dropout layers, and dense layers. Early stopping and validation data are utilized to prevent overfitting. Intel's OpenVINO toolkit is instrumental in optimizing the model for efficient inference on Intel hardware.
Technologies Used
- TensorFlow and Keras for deep learning
- Python, NumPy, Pandas, and Matplotlib for data manipulation and visualization
- OpenVINO for model optimization
- Streamlit for deployment
- AWS EC2 for hosting
- Git/GitHub for version control
- Intel CPUs and hardware acceleration for improved performance
Documents and Presentations
Repository
https://github.com/DakshDadhania/Intel-Unnati_BrainWaves