Google Landmark Recognition 2020 Competition Third Place Solution
This work addresses landmark recognition for computer vision applications, but it is incremental as it builds on existing methods for a competition.
The paper tackled the Google Landmark Recognition 2020 competition by developing an ensemble of Sub-center ArcFace models with dynamic margins to handle dataset imbalance, achieving third place with scores of 0.6344 and 0.6289 on the private leaderboard.
We present our third place solution to the Google Landmark Recognition 2020 competition. It is an ensemble of global features only Sub-center ArcFace models. We introduce dynamic margins for ArcFace loss, a family of tune-able margin functions of class size, designed to deal with the extreme imbalance in GLDv2 dataset. Progressive finetuning and careful postprocessing are also key to the solution. Our two submissions scored 0.6344 and 0.6289 on private leaderboard, both ranking third place out of 736 teams.