AI Paper Atlas

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

Simpler Is Better: Few-Shot Semantic Segmentation With Classifier Weight Transformer

Zhihe Lu, Sen He, Xiatian Zhu, Li Zhang, Yi-Zhe Song, Tao Xiang

ICCV 2021 · Detection and segmentation › Open-set and few-shot detection · approach: Few-shot segmentation · 224 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
meta-train the classifier only instead of the whole model (ablation)CWT · COCO-20i 1-shot (ResNet-50)mIoU (%): 21.5 -> 32.9Improved
add CWT classifier adaptation to each query image on top of pre-trained classifier (ablation)CWT · COCO-20i 1-shot (ResNet-50)mIoU (%): 28.6 -> 32.9Improved
condition CWT on the query image instead of the support image (ablation)CWT · COCO-20i 1-shot (ResNet-50)mIoU (%): 18.2 -> 32.9Improved
pre-train encoder/decoder and train classifier on support only (no meta-learning) vs PPNet (ablation)pre-training baseline · COCO-20i 1-shot (ResNet-50)mIoU (%): 25.7 -> 28.6Improved
use CWT vs best competitor with ResNet-50CWT · COCO-20i 1-shotmIoU (%): 30.6 -> 32.9Improved
use CWT vs PPNet in 5-shot with ResNet-50CWT · COCO-20i 5-shotmIoU (%): 36.2 -> 41.3Improved
use CWT vs PFENet in 5-shot with ResNet-101CWT · COCO-20i 5-shotmIoU (%): 37.4 -> 42.0Improved
use CWT vs PFENet in 1-shotCWT · PASCAL-5i 1-shot (ResNet-50)mIoU (%): 60.8 -> 56.4Hurt
use CWT vs PFENet in 5-shotCWT · PASCAL-5i 5-shot (ResNet-50)mIoU (%): 61.9 -> 63.7Improved
use CWT vs RPMMs in COCO to PASCAL transferCWT · COCO to PASCAL 1-shotmIoU (%): 49.6 -> 59.5Improved
use CWT vs RPMMs in PASCAL to COCO transferCWT · PASCAL to COCO 5-shotmIoU (%): 41.0 -> 45.8Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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