Query Optimization Detection Transformer for Small Objects in Remote Sensing Images
Keywords: Remote sensing images; Small object detection; Transformer; Query optimization
TL;DR: A detection Transformer for small objects in remote sensing images is proposed, called QO-DETR.
Abstract: Object detection in remote sensing images is a challenging task. Remote sensing images contain substantial background noise and complex contextual information, which weakens the feature representation of small objects, making detection difficult. To solve these problems, a detection Transformer for small objects in remote sensing images is proposed, called QO-DETR. Specifically, to enhance the feature representation of small objects, a query proposal generation module is designed to select queries based on multi-class classification scores. These queries provide the initial position embeddings for object queries in the decoder, enabling the decoder's attention mechanism to focus on object regions. To improve the model’s robustness to noise, a group denoising module is designed to add noise into decoder queries during training, enhancing the network's ability to reconstruct object features from noise. To accurately locate small objects, a query cascade refinement strategy is designed, and each decoder layer refines anchor parameters under the guidance of preceding layers to achieve spatial alignment between the anchor and the object. Experiments have been carried out on DIOR and AI-TOD. The AP and APs on DIOR reach 51.3% and 13.4%, respectively, while on AI-TOD, they reach 23.6% and 30.1%. QO-DETR shows superior performance in detecting small objects.
Primary Area: applications to computer vision, audio, language, and other modalities
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 2866
Loading