Layer-Wise Query Selection to Eliminate Redundant Queries in DETR
Open Access
- 9 July 2025
- journal article
- research article
- Published by MDPI AG in Applied Sciences
- Vol. 15 (14) , 7686
Abstract
Recent advancements in the detection Transformer model have demonstrated remarkable accuracy in real-time object detection using an end-to-end approach. DETR leverages the concept of object queries, which act as “questions” to determine the presence and location of objects. However, the excessive number of object queries significantly increases computational complexity, leading to higher training and inference times, greater memory consumption, and increased costs. To address this issue, this paper introduces the selected query detection Transformer, a novel approach that optimizes the selection of object queries in the decoder. By progressively filtering and reducing unnecessary queries, the selected query detection Transformer maintains model performance while significantly reducing computational overhead, resulting in faster training and inference times.Keywords
Funding Information
- Defense Acquisition Program Administration (DAPA) (915073201, RS-2025-00520064)
- Korea Government (MSIT) (915073201, RS-2025-00520064)
This publication has 22 references indexed in Scilit:
- Dense Distinct Query for End-to-End Object Detection2023
- DN-DETR: Accelerate DETR Training by Introducing Query DeNoising2022
- Dynamic DETR: End-to-End Object Detection with Dynamic Attention2021
- Fast Convergence of DETR with Spatially Modulated Co-Attention2021
- End-to-End Object Detection with Transformers2020
- Path Aggregation Network for Instance Segmentation2018
- SSD: Single Shot MultiBox Detector2016
- Deep Residual Learning for Image Recognition2016
- You Only Look Once: Unified, Real-Time Object Detection2016
- Microsoft COCO: Common Objects in Context2014