AI Paper Atlas

Research map / Detection and segmentation / CNN features

OS2D: One-Stage One-Shot Object Detection by Matching Anchor Features

Anton Osokin, Denis Sumin, Vasily Lomakin

ECCV 2020 · Detection and segmentation › CNN features · approach: Box regression and loss design · 54 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 RLL loss instead of contrastive loss (ablation)OS2D V1 · GroZi-3.2k val-new-clmAP: 81.8 -> 85.7Improved
add recognition target remapping (ablation)OS2D V1 · GroZi-3.2k val-new-clmAP: 85.7 -> 87.0Improved
add hard patch mining (ablation)OS2D V1 · GroZi-3.2k val-new-clmAP: 87.0 -> 86.5No clear effect
add hard patch mining (ablation)OS2D V2 · GroZi-3.2k val-new-clmAP: 89.5 -> 90.6Improved
initialize TransformNet from Rocco et al. weights (init transform) (ablation)OS2D V2 RLL + remap · GroZi-3.2k val-new-clmAP: 87.0 -> 88.0Improved
zero out the localization loss (ablation)OS2D V2 · GroZi-3.2k val-new-clmAP: 88.0 -> 89.5Improved
use V2 (full affine, weak supervision) instead of V1 TransformNetOS2D · GroZi-3.2k val-new-clmAP: 88.7 -> 90.6Improved
pretrain feature extractor on ImageNet classification instead of random init (ablation)OS2D V2-train · GroZi-3.2k val-new-clmAP: 67.9 -> 89.0Improved
initialize from COCO detection instead of ImageNet classification (ablation)OS2D V2-train · GroZi-3.2k val-new-clmAP: 89.5 -> 88.1Hurt
replace batch norm with group norm in the converted ImageNet model (ablation)OS2D V2-train · GroZi-3.2k val-new-clmAP: 89.5 -> 87.1Hurt
add learned transformation model (vs sliding window with target aspect ratio) (ablation)OS2D · GroZi-3.2k paste-vmAP: 65.1 -> 73.3Improved
use one-shot matching OS2D instead of detector plus retrievalOS2D V2-train · GroZi-3.2k paste-vmAP: 44.3 -> 73.3Improved

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