Relation-enhanced DETR for Component Detection in Graphic Design Reverse Engineering

Link
Abstract

It is a common practice for designers to create digital prototypes from a mock-up/screenshot. Reverse engineering graphic design by detecting its components (e.g., text, icon, button) helps expedite this process. This paper first conducts statistical analysis to emphasize the importance of relations in graphic layouts, which further motivates us to incorporate relation modeling into component detection. Built on the current state-of-the-art DETR (DEtection TRansformer), we introduce a learnable relation matrix to model class correlations. Specifically, the matrix will be added to the DETR decoder to update the query-to-query self-attention. Experiment results on three public datasets show that our approach achieves better performance than several strong baselines. We further visualize the learned relation matrix and observe some reasonable patterns. Moreover, we show an application of component detection where we leverage the detection outputs as augmented training data for layout generation, which achieves promising results.

Synth

Problem:: 레이아웃 검출에서 각 BBox간의 관계 모델링이 중요함에도 적용되지 않음

Solution:: 관계를 모델링 할 수 있는 요소를 추가

Novelty:: 디자인 레이아웃에서 관계의 중요성을 분석/DETR의 Self-Attention에 Relation 정보를 최초로 적용

Note:: 실험 결과가 미심쩍음. DETR이 예상보다 너무 검출을 못함

Summary

Motivation

file-20250318180626645.png|500

Method

file-20250318181354267.png|850

검증 방법