AI Paper Atlas

Research map / Detection and segmentation / 3D detection and driving

Unsupervised Domain Adaptation for Monocular 3D Object Detection via Self-Training

Zhenyu Li, Zehui Chen, Ang Li, Liangji Fang, Qinhong Jiang, Xianming Liu, Junjun Jiang

ECCV 2022 · Detection and segmentation › 3D detection and driving · approach: Monocular 3D object detection · 35 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 geometry-aligned multi-scale training (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 0 -> 21.89Improved
add GAMS to naive self-training (ablation)Naive ST · NuScenes->KITTIAP40 3D Moderate: 0 -> 3.69Improved
use end-to-end mean teacher instead of naive ST (both with GAMS) (ablation)STMono3D · NuScenes->KITTIAP40 3D Moderate: 3.69 -> 16.85Improved
add EMA teacher (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 0.82 -> 21.89Improved
add QAS reweighting on classification loss (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 18.14 -> 21.89Improved
add QAS reweighting on regression loss (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 18.14 -> 13.80Hurt
apply QAS to both cls and reg losses (vs cls only) (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 21.89 -> 15.67Hurt
add positive focusing training (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 7.80 -> 13.15Improved
add dynamic threshold (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 7.80 -> 13.73Improved
combine PFT and DT (ablation)STMono3D · NuScenes->KITTIAP11 3D Moderate: 7.80 -> 21.89Improved
train with STMono3D vs oracle source-trained modelSTMono3D · NuScenes->KITTIAP11 3D Moderate: 19.88 -> 21.89Improved
train with STMono3D vs oracleSTMono3D · NuScenes->KITTIAP40 3D Moderate: 18.97 -> 16.85Hurt

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