AI Paper Atlas

Research map / Detection and segmentation / CNN features

Measuring the Impact of Rotation Equivariance on Aerial Object Detection

Xiuyu Wu, Xinhao Wang, Xiubin Zhu, Lan Yang, Jiyuan Liu, Xingchen Hu

ICCV 2025 · Detection and segmentation › CNN features · approach: Box regression and loss design · 8 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use strictly rotation-equivariant downsampling (ablation)MessDet (no head) · DOTA-v1.0mAP: 78.15 -> 78.51Improved
use strict equivariant downsampling (ablation)RTMDet (no EMA) · DOTA-v1.0mAP: 77.34 -> 77.10No clear effect
use strictly equivariant downsampling (ablation)MessDet (no head) · DOTA-v1.5mAP: 71.26 -> 72.42Improved
add RE-CA to strict MessDet (ablation)Str. MessDet · DOTA-v1.0mAP: 76.91 -> 78.51Improved
add RE-CA to approximate MessDet (ablation)Appr. MessDet · DOTA-v1.0mAP: 77.47 -> 78.15Improved
use multi-branch head with 3 conv modules instead of RTMDet head (ablation)Appr. MessDet · DOTA-v1.0mAP: 78.15 -> 78.45Improved
use multi-branch head with 5 conv modules (ablation)Appr. MessDet · DOTA-v1.0mAP: 78.15 -> 77.62Hurt
use multi-branch head with 2 conv modules (ablation)Appr. MessDet · DOTA-v1.0mAP: 78.15 -> 77.46Hurt
strict vs approximate MessDetMessDet · DOTA-v1.0 single-scalemAP: 78.45 -> 79.12Improved
strict vs approximate MessDetMessDet · DOTA-v1.5mAP: 72.38 -> 73.14Improved
strict vs approximate MessDetMessDet · DIOR-RmAP: 67.42 -> 68.19Improved
strict vs approximate MessDetMessDet · DOTA-v1.0 multi-scalemAP: 80.36 -> 81.07Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy