Skip to content

README.md

Latest commit

 

History

History
58 lines (47 loc) · 2.46 KB

README.md

File metadata and controls

58 lines (47 loc) · 2.46 KB
Raw

Content-Based Image Retrieval (CBIR) Framework

Overview

This project presents a Hybrid Deep Hashing and Metric Space Partitioning Framework for scalable Content-Based Image Retrieval (CBIR). It integrates unsupervised representation learning with VP-Tree optimization to enhance retrieval accuracy and efficiency.

Features

  • Autoencoder-Based Feature Extraction: Generates compact and semantically meaningful image representations.
  • Hybrid Search Mechanism: Uses Euclidean-based nearest neighbor search (O(N log N)) for moderate-scale datasets and VP-Tree-based indexing (O(log N)) for large-scale retrieval.
  • Deep Hashing Techniques: Enhances retrieval precision while maintaining computational efficiency.
  • Scalability and Speed Optimization: Reduces search complexity without compromising retrieval accuracy.

Installation

Prerequisites

Ensure you have the following installed:

  • Python 3.8+
  • PyTorch
  • NumPy
  • OpenCV
  • Matplotlib
  • scikit-learn

Setup

Clone the repository and install dependencies:

git clone https://github.com/rapexa/Scalable-Content-Based-Image-Retrieval.git
cd Scalable-Content-Based-Image-Retrieval
pip install -r requirements.txt
jupyter lab .

Results

The proposed model achieves higher mean average precision (mAP), reduced search latency, and improved storage efficiency compared to traditional CBIR techniques. Benchmark results on CIFAR-10 and ImageNet datasets demonstrate its superior performance.The experimental results validate the superiority of the proposed method, outperforming traditional CBIR techniques in terms of retrieval speed, accuracy, and computational efficiency. The proposed model achieves:

  • 96.1% mAP
  • 94.8% F1-Score
  • 40% reduction in retrieval time compared to traditional models.

Conclusion

This research provides a novel approach to CBIR systems, combining deep feature learning, adaptive hashing, and efficient indexing structures to deliver scalable, real-time image retrieval solutions for large datasets.

Paper & Documentation

For an in-depth explanation, refer to the research paper: Download Paper (PDF)

Citation

If you use this work, please cite:

@article{your_paper,
  title={A Hybrid Deep Hashing and Metric Space Partitioning Framework for Scalable Content-Based Image Retrieval},
  author={S. Mohamadzadeh, M. Gharehbagh},
  journal={Your Journal},
  year={2025}
}

License

This project is licensed under the MIT License.