Large-scale Landmark Retrieval/Recognition under a Noisy and Diverse Dataset

TL;DR

This paper introduces a robust landmark retrieval and recognition system designed for large, noisy, and diverse datasets, utilizing deep neural networks, metric learning, automated data cleaning, and re-ranking, achieving top results in major challenges.

Contribution

The authors develop a novel system combining deep metric learning with automated data cleaning and re-ranking to improve landmark recognition in noisy, diverse datasets.

Findings

Achieved 1st place in Google Landmark Retrieval 2019 challenge.

Secured 3rd place in Google Landmark Recognition 2019 challenge.

Demonstrated robustness to noise and diversity in large-scale datasets.

Abstract

The Google-Landmarks-v2 dataset is the biggest worldwide landmarks dataset characterized by a large magnitude of noisiness and diversity. We present a novel landmark retrieval/recognition system, robust to a noisy and diverse dataset, by our team, smlyaka. Our approach is based on deep convolutional neural networks with metric learning, trained by cosine-softmax based losses. Deep metric learning methods are usually sensitive to noise, and it could hinder to learn a reliable metric. To address this issue, we develop an automated data cleaning system. Besides, we devise a discriminative re-ranking method to address the diversity of the dataset for landmark retrieval. Using our methods, we achieved 1st place in the Google Landmark Retrieval 2019 challenge and 3rd place in the Google Landmark Recognition 2019 challenge on Kaggle.