• Home
  • Search
  • Exploring Sequence Feature Alignment for Domain Adaptive Detection Transformers
  • Open Access IconOpen Access
  • Cite Icon103
  • https://doi.org/10.1145/3474085.3475317Copy DOI Icon

Exploring Sequence Feature Alignment for Domain Adaptive Detection Transformers

  • Oct 17, 2021
  • Wen Wang +6 more
Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

Detection transformers have recently shown promising object detection results\nand attracted increasing attention. However, how to develop effective domain\nadaptation techniques to improve its cross-domain performance remains\nunexplored and unclear. In this paper, we delve into this topic and empirically\nfind that direct feature distribution alignment on the CNN backbone only brings\nlimited improvements, as it does not guarantee domain-invariant sequence\nfeatures in the transformer for prediction. To address this issue, we propose a\nnovel Sequence Feature Alignment (SFA) method that is specially designed for\nthe adaptation of detection transformers. Technically, SFA consists of a domain\nquery-based feature alignment (DQFA) module and a token-wise feature alignment\n(TDA) module. In DQFA, a novel domain query is used to aggregate and align\nglobal context from the token sequence of both domains. DQFA reduces the domain\ndiscrepancy in global feature representations and object relations when\ndeploying in the transformer encoder and decoder, respectively. Meanwhile, TDA\naligns token features in the sequence from both domains, which reduces the\ndomain gaps in local and instance-level feature representations in the\ntransformer encoder and decoder, respectively. Besides, a novel bipartite\nmatching consistency loss is proposed to enhance the feature discriminability\nfor robust object detection. Experiments on three challenging benchmarks show\nthat SFA outperforms state-of-the-art domain adaptive object detection methods.\nCode has been made available at: https://github.com/encounter1997/SFA.\n

Similar Papers
  • PDF
  • Research Article
  • Citations22

Multi-Scale Feature Enhanced Domain Adaptive Object Detection For Power Transmission Line Inspection

  • Jan 01, 2020
  • IEEE Access
  • Pengyu Zhang +5
  • Research Article
  • Citations2

DAPoinTr: Domain Adaptive Point Transformer for Point Cloud Completion

  • Apr 11, 2025
  • Proceedings of the AAAI Conference on Artificial Intelligence
  • Yinghui Li +6
  • Book Chapter
  • Citations3

Foreground Feature Selection and Alignment for Adaptive Object Detection

  • Jan 01, 2021
  • Zixuan Huang +2
  • Research Article
  • Citations4

FIE-Net: Foreground Instance Enhancement Network for Domain Adaptation Object Detection in Remote Sensing Imagery

  • Jan 01, 2025
  • IEEE Transactions on Geoscience and Remote Sensing
  • Jun Zhang +4
  • Research Article
  • Citations30

Source data-free domain adaptation for a faster R-CNN

  • Nov 19, 2021
  • Pattern Recognition
  • Lin Xiong +4
  • Research Article
  • Citations5

Loose to compact feature alignment for domain adaptive object detection

  • Mar 27, 2024
  • Pattern Recognition Letters
  • Yang Li +3
  • Conference Article
  • Citations2

SDAN-YOLO:Self-attention domain adaptive network for YOLOv7

  • Dec 17, 2023
  • Jian Zhang +3
  • Research Article
  • Citations21

Rethinking Local and Global Feature Representation for Dense Prediction

  • Nov 12, 2022
  • Pattern Recognition
  • Mohan Chen +4
  • Research Article
  • Citations1

VLSDA: Vision–Language Model-Supervised Domain Adaptation for Cross-Domain Object Detection in Remote Sensing

  • Jan 01, 2025
  • IEEE Transactions on Geoscience and Remote Sensing
  • Junhong Lu +1
  • Conference Article
  • Citations2

Integrated Generative-Model Domain-Adaptation for Object Detection under Challenging Conditions

  • Jun 01, 2022
  • Mazin Hnewa +1
  • Research Article
  • Citations5

A dynamically class-wise weighting mechanism for unsupervised cross-domain object detection under universal scenarios

  • Jun 03, 2024
  • Knowledge-Based Systems
  • Wenxu Shi +3
  • Research Article
  • Citations13

Global and local attentional feature alignment for domain adaptive nuclei detection in histopathology images

  • Jul 02, 2022
  • Artificial Intelligence in Medicine
  • Zhi Wang +5
  • Research Article
  • Citations5

Adaptive feature alignment network with noise suppression for cross-domain object detection

  • Oct 28, 2024
  • Neurocomputing
  • Wei Jiang +6
  • Conference Article
  • Citations9

Using Feature Alignment Can Improve Clean Average Precision And Adversarial Robustness In Object Detection

  • Sep 19, 2021
  • Weipeng Xu +2
  • Research Article
  • Citations34

Joint feature representation learning and progressive distribution matching for cross-project defect prediction

  • Apr 07, 2021
  • Information and Software Technology
  • Quanyi Zou +4
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.