AI Paper Atlas

Research map / Detection and segmentation / CNN features

SCRDet++: Detecting Small, Cluttered and Rotated Objects via Instance-Level Feature Denoising and Rotation Loss Smoothing

Xue Yang, Junchi Yan, Wenlong Liao, Xiaokang Yang, Jin Tang, Tao He

TPAMI 2022 · Detection and segmentation › CNN features · approach: Box regression and loss design · 390 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
add binary-mask InLD supervision (ablation)R3Det · DOTA-v1.0 OBBmAP: 65.73 -> 68.12Improved
use multi-class instead of binary mask in InLD (ablation)R3Det · DOTA-v1.0 OBBmAP: 68.12 -> 69.43Improved
multiply objectness P(object) into class score (coproduct) (ablation)R3Det InLD · DOTA-v1.0 OBBmAP: 69.43 -> 69.81Improved
add InLD module (ablation)R3Det · DOTA-v1.0 OBBFPS: 14 -> 13Hurt
add InLD module (ablation)RetinaNet-H · DOTA-v1.0 OBBmAP: 62.21 -> 65.40Improved
add InLD module (ablation)FPN · DIORmAP: 71.74 -> 73.21Improved
add InLD module (ablation)RetinaNet · COCOmAP: 34.4 -> 35.8Improved
add InLD module (ablation)RetinaNet · ICDAR2015F-measure: 77.13 -> 78.68Improved
add InLD module (ablation)RetinaNet · S2TLDmAP: 93.25 -> 94.11Improved
add InLD module (ablation)FPN · BSTLDmAP: 67.46 -> 68.67Improved
use dilated convs without the InLD loss (ablation)R3Det · DOTA-v1.0 OBBmAP: 65.73 -> 66.62No clear effect
add ImLD (nonlocal gaussian) image-level denoising (ablation)R3Det · DOTA-v1.0 OBBmAP: 65.73 -> 67.68Improved

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