Research image deduplication

[IvanHristov98]

Research image deduplication methods and implement them as deduplication use cases* in the project.

Benchmark each researched method for accuracy, speed and memory.

• use cases comes from clean architecture.

---

Papers:

• [x] Large Scale Image Deduplication by Tzay-Yeu Wen
• [x] SIFT by David G. Lowe
• [x] Robust Wide Baseline Stereo from Maximally Stable Extremal Regions
• [x] Visual Categorization with Bags of Keypoints
• [x] Scalable Recognition with a Vocabulary Tree
• [x] Near Duplicate Image Detection: min-Hash and tf-idf Weighting
• [x] Bundling Features for Large Scale Partial-Duplicate Web Image Search
• [x] Aggregating local descriptors into a compact image representation

[IvanHristov98]

I went through a series of videos explaining SIFT from Columbia university. Furthermore I read the paper introducing SIFT and then tried to implement it.

It turned out that SIFT is already made available by opencv and is no longer under a patent. Hence we could freely use it.

A simple implementation comparing two images and finding whether they are duplicate is available in this project.