2nd Place Solution To Facebook AI Image Similarity Challenge Matching Track

Abstract

This paper presents the 2nd place solution to the Facebook AI Image Similarity Challenge : Matching Track on DrivenData. The solution is based on self-supervised learning, and Vision Transformer(ViT). The main breaktrough comes from concatenating query and reference image to form as one image and asking ViT to directly predict from the image if query image used reference image. The solution scored 0.8291 Micro-average Precision on the private leaderboard.

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyjeon20212nd

Related papers

• 3rd Place: A Global And Local Dual Retrieval Solution To Facebook AI Image Similarity Challenge (2021)0.00
• Google Landmark Retrieval 2021 Competition Third Place Solution (2021)0.00
• 3rd Place Solution To "google Landmark Retrieval 2020" (2020)0.00
• 2nd Place Solution To Google Landmark Retrieval 2021 (2021)4.74
• 1st Place Solution To Google Landmark Retrieval 2020 (2020)0.00
• QK Iteration: A Self-supervised Representation Learning Algorithm For Image Similarity (2021)0.00
• Boosting Vision Transformers For Image Retrieval (2022)15.28
• 2nd Place Solution To Google Landmark Retrieval 2020 (2022)0.00