AI Paper Atlas

Research map / Detection and segmentation / CNN features

SCRDet: Towards More Robust Detection for Small, Cluttered and Rotated Objects

Xue Yang, Jirui Yang, Junchi Yan, Yue Zhang, Tengfei Zhang, Zhi Guo, Xian Sun, Kun Fu

ICCV 2019 · Detection and segmentation › CNN features · approach: Box regression and loss design · 1,111 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 pixel attention (ablation)R2CNN · DOTA OBBmAP (%): 60.67 -> 64.34Improved
add channel attention to pixel attention (MDA-Net) (ablation)R2CNN+Pixel Attention · DOTA OBBmAP (%): 64.34 -> 65.33Improved
add SF-Net (finer sampling and feature fusion) (ablation)R2CNN+MDA · DOTA OBBmAP (%): 65.33 -> 68.89Improved
add IoU-smooth L1 loss (ablation)R2CNN+SF+MDA · DOTA OBBmAP (%): 68.89 -> 69.83Improved
add image pyramid training and testing (ablation)R2CNN+SF+MDA+IoU · DOTA OBBmAP (%): 69.83 -> 72.61Improved
add supervision to the attention network (ablation)MDA-Net · DOTA OBBmAP (%): 61.23 -> 65.33Improved
add supervision to the attention network (ablation)MDA-Net · VOC2007 (07+12 train)mAP (%): 80.53 -> 82.27Improved
add shifted anchors (SA) (ablation)R2CNN+MDA · DOTA OBBmAP (%): 65.33 -> 62.78Hurt
add shift jittering (SJ) (ablation)R2CNN+MDA · DOTA OBBmAP (%): 65.33 -> 62.70Hurt
use dilated convolution to enlarge feature maps (ablation)R2CNN+MDA · DOTA OBBmAP (%): 65.33 -> 66.40Improved
reduce anchor stride from 16 to 6 (ablation)SCRDet · DOTA OBBmAP (%): 63.64 -> 67.06Improved
add MDA-NetFPN* · COCO minivalmAP/F1: 36.1 -> 36.8Improved

Later papers that compared with this method

6 later accepted papers report SCRDet as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.

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