AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

The Devil is in Classification: A Simple Framework for Long-tail Instance Segmentation

Tao Wang, Yu Li, Bingyi Kang, Junnan Li, Junhao Liew, Sheng Tang, Steven Hoi, Jiashi Feng

ECCV 2020 · Detection and segmentation › Open-set and few-shot detection · approach: Amodal and open-world instance segmentation · 186 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 classification-head calibration with bi-level sampling (ablation)Mask R-CNN r50-ag · LVIS v0.5 valmask AP1: 0.0 -> 13.2Improved
add classification-head calibration with bi-level sampling (ablation)Mask R-CNN r50 (class-wise) · LVIS v0.5 valmask AP4: 29.6 -> 25.3Hurt
add dual head inference on top of calibration (ablation)Mask R-CNN r50 · LVIS v0.5 valmask AP: 22.3 -> 23.4Improved
add dual head inference on top of calibration (ablation)Mask R-CNN r50-ag · LVIS v0.5 valmask AP: 21.2 -> 22.2Improved
combine heads by selecting by class instance count instead of averaging predictions (ablation)SimCal dual head · LVIS v0.5 valmask AP: 20.3 -> 21.1Improved
initialize calibration head from the original 3-layer head instead of random initialization (ablation)SimCal head · LVIS v0.5 val (HTC)mask AP1 / AP2: +4.1 AP1, +4.3 AP2Improved
vary head/tail boundary T from 10 to 1000 (ablation)SimCal dual head · LVIS v0.5 valAP: T 90 to 500 near optimal (T = 300)No clear effect
use class-agnostic heads instead of class-wise heads with calibration and dual head (ablation)SimCal · LVIS v0.5 valmask AP1: 10.2 -> 13.3Improved
add loss re-weighting during trainingMask R-CNN r50 · LVIS v0.5 valmask AP4: 29.6 -> 25.1Hurt
add image-level repeat sampling during trainingMask R-CNN r50 · LVIS v0.5 valmask AP: 20.7 -> 23.2Improved
add decoupled calibration and dual head to existing loss methods CM, LR, FL (ablation)Mask R-CNN r50 · LVIS v0.5 valmask AP1: 2.6 -> 4.6; 6.0 -> 6.9; 6.2 -> 7.1Improved
add decoupled calibration to image-level repeat sampling (ablation)Mask R-CNN r50 · LVIS v0.5 valmask AP: 23.2 -> 22.7Hurt

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