ICME 2022 Few-shot LOGO detection top 9 solution

TL;DR

This paper presents a detailed account of techniques used in a few-shot logo detection competition, highlighting methods to handle tiny logos, similar brands, and adversarial images with limited annotations, achieving top 10 rankings.

Contribution

The paper introduces specific strategies and techniques tailored for few-shot logo detection in challenging scenarios, contributing practical solutions for real-world logo recognition tasks.

Findings

Achieved top 10 rankings in the competition

Developed methods for tiny logo detection

Handled similar brands and adversarial images effectively

Abstract

ICME-2022 few-shot logo detection competition is held in May, 2022. Participants are required to develop a single model to detect logos by handling tiny logo instances, similar brands, and adversarial images at the same time, with limited annotations. Our team achieved rank 16 and 11 in the first and second round of the competition respectively, with a final rank of 9th. This technical report summarized our major techniques used in this competitions, and potential improvement.