AI Paper Atlas

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

Unsupervised Prompt Tuning for Text-Driven Object Detection

Weizhen He, Weijie Chen, Binbin Chen, Shicai Yang, Di Xie, Luojun Lin, Donglian Qi, Yueting Zhuang

ICCV 2023 · Detection and segmentation › Open-set and few-shot detection · approach: Open-vocabulary object detection · 9 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 Dual Complementary Teaching with offline and online teachers (ablation)UPT · CityscapesmAP50: 40.0 -> 42.4Improved
add Nested Mean Teaching on top of Dual Complementary Teaching (ablation)UPT · CityscapesmAP50: 42.4 -> 43.1Improved
feed strongly augmented images to online teacher (ablation)Dual Complementary Teaching · CityscapesmAP50: 40.5 -> 42.4Improved
use strong instead of weak augmentation for offline teacher (ablation)Dual Complementary Teaching · CityscapesmAP50: 42.4 -> 41.7Hurt
increase inner-loop steps K from 0 to 3 (ablation)Nested Mean Teaching · CityscapesmAP50: 42.4 -> 43.1Improved
increase inner-loop steps K from 3 to 4 (ablation)Nested Mean Teaching · CityscapesmAP50: 43.1 -> 42.7Hurt
tune entire model instead of text prompt without labels (ablation)GLIP-T unsupervised tuning · CityscapesmAP50: 33.8 -> 12.8Hurt
tune only visual encoder instead of text prompt without labels (ablation)GLIP-T unsupervised tuning · CityscapesmAP50: 33.8 -> 26.5Hurt
unsupervised prompt tuning vs conventional Unbiased Teacher mean teachingGLIP-T · CityscapesmAP50: 40.0 -> 43.1Improved
apply UPT to datasets with small zero-shot-to-oracle gapGLIP-T · MS-COCOmAP50: 63.5 -> 64.1No clear effect
apply Unbiased Teacher on single-class datasetsGLIP-T · KITTI/Sim10K/EgoHands/Pistols averagemAP50: 68.5 -> 67.6Hurt
UPT without labels vs supervised few-shot prompt tuningGLIP-T · CityscapesmAP50: 43.1 vs 42.5 (5-shot)Improved

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