This repository is a curated collection of academic and industry papers on Content-Based Image Retrieval (CBIR), targeting researchers and engineers in computer vision and machine learning. It provides a structured overview of seminal works and recent advancements across various CBIR sub-fields, serving as a valuable resource for understanding the landscape and identifying key contributions.

How It Works

The collection is organized thematically, categorizing papers by their core technical contributions. This includes classical local features (e.g., SIFT, VLAD), deep learning-based global and local features, instance search techniques, approximate nearest neighbor (ANN) search methods, and applications in industry. This structured approach allows users to navigate the field efficiently, from foundational concepts to state-of-the-art deep learning models.

Highlighted Details

• Comprehensive coverage of classical and deep learning-based feature extraction methods.
• Detailed sections on Approximate Nearest Neighbor (ANN) search libraries and techniques.
• Inclusion of papers related to CBIR attacks, ranking, and industry applications (e.g., Bing, Pinterest, Alibaba).
• Links to associated code, datasets, and benchmarks for many listed papers.

Maintenance & Community

This is a static collection of papers, with no active development or community interaction channels mentioned. The primary contributor is willard-yuan.

Licensing & Compatibility

The repository itself does not host code or data, and therefore does not have a specific license. The linked papers are subject to their respective publication licenses.

Limitations & Caveats

This is a curated list of papers and does not provide any implementation, code, or direct tooling for CBIR. The content is purely informational and relies on external resources for practical application.