Google Landmark Retrieval Challenge
Organised by: Bhavan Chahal & Chris Norman (two groups, same problem)
Abstract:
Image retrieval is a fundamental problem in computer vision: given a query image, can you find similar images in a large database? This is especially important for query images containing landmarks, which accounts for a large portion of what people like to photograph.
In this Kaggle competition, competitors are given query images and, for each query, are expected to retrieve all database images containing the same landmarks (if any).
The new dataset is the largest worldwide dataset for image retrieval research, comprising more than a million images of 15K unique landmarks. We hope that this release will accelerate progress in this important research problem.
If we decide to submit, submissions are evaluated according to mean Average Precision, and the prizes are 1st place - $ 1,250, 2nd place - $ 750, and 3rd place - $ 500. The deadline is May 15, 2018, a few weeks after the retreat.
Aim:
Given an image, can you find all of the same landmarks in a dataset? We would like to have two groups approaching the problem differently and see which performs best!
Objectives:
- Analyse the data and format as required
- Attempt the Kaggle competition
- Have fun with data science!
Of Interest to:
Google, data scientists, and those interested in vision tasks
Resources Necessary:
- Competition data (Available here: https://www.kaggle.com/c/landmark-retrieval-challenge/data)
- R or Python and relevant packages
References:
- Kaggle competition page: https://www.kaggle.com/c/landmark-retrieval-challenge
- H. Noh, A. Araujo, J. Sim, T. Weyand, B. Han, "Large-Scale Image Retrieval with Attentive Deep Local Features", Proc. ICCV'17