AI Paper Atlas

Research map / Detection and segmentation / CNN features

Bootstrap Your Object Detector via Mixed Training

Mengde Xu, Zheng Zhang, Fangyun Wei, Yutong Lin, Yue Cao, Stephen Lin, Han Hu, Xiang Bai

NeurIPS 2021 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 0 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
replace single-input training with MixTrainingFaster R-CNN ResNet-50 · COCO2017 valmAP: 41.7 -> 44.0Improved
replace SiTraining with MixTraining on Swin-S backboneFaster R-CNN Swin-S · COCO2017 valmAP: 48.7 -> 50.3Improved
replace SiTraining with MixTrainingCascade R-CNN Swin-S · COCO2017 valmAP: 50.9 -> 52.8Improved
extend training from 360K to 720K iterations with MixTrainingFaster R-CNN ResNet-50 · COCO2017 valmAP: 42.4 -> 44.0Improved
extend SiTraining (normal aug) from 360K to 720K iterationsFaster R-CNN ResNet-50 · COCO2017 valmAP: 41.7 -> 36.1Hurt
use strong augmentation in SiTraining and extend to 720KFaster R-CNN ResNet-50 · COCO2017 valmAP: 40.1 -> 38.9Mixed
add mixed data augmentation (ablation)SiTraining · COCO2017 valmAP: 41.7 -> 42.5Improved
add mixed training targets (pseudo boxes) on top of mixed augmentation (ablation)MixTraining · COCO2017 valmAP: 42.5 -> 44.0Improved
add pseudo boxes for missing labels (ablation)MixTraining · COCO2017 valmAP: 42.5 -> 43.7Improved
add pseudo boxes for localization noise only (ablation)MixTraining · COCO2017 valmAP: 42.5 -> 42.9Improved
use hybrid of both pseudo box strategies instead of missing label only (ablation)MixTraining · COCO2017 valmAP: 43.7 -> 44.0Improved
apply mixed augmentation strategy without weighted loss (ablation)SiTraining · COCO2017 valmAP: 41.7 -> 42.1Improved

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