AI Paper Atlas

Research map / Detection and segmentation / CNN features

Dynamic Token Pruning in Plain Vision Transformers for Semantic Segmentation

Quan Tang, Bowen Zhang, Jiajun Liu, Fagui Liu, Yifan Liu

ICCV 2023 · Detection and segmentation › CNN features · approach: Plain and high-resolution ViTs · 68 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
finetune heads for 40k iterations with DToP instead of applying directly (ablation)DToP · ADE20KmIoU (%): 47.9 -> 49.8Improved
retrain the entire model with DToP instead of finetuning heads (ablation)DToP · ADE20KmIoU (%): 49.8 -> 49.1Hurt
keep top-k tokens per category when pruning (ablation)DToP · ADE20KmIoU (%): 48.7 -> 49.5Improved
combine average and top-k context tokens (ablation)DToP · ADE20KmIoU (%): 49.5 -> 49.7Improved
lower confidence threshold p0 from 0.95 to 0.60 (ablation)DToP with ATM head · ADE20KmIoU (%): 49.8 -> 46.8Hurt
split backbone into 4 stages instead of 3 (ablation)DToP · ADE20KmIoU (%): 49.8 -> 48.3Hurt
use a powerful ATM auxiliary head instead of FCN for grading difficulty (ablation)DToP · ADE20KmIoU (%): 46.9 -> 48.4Improved
apply DToP@Finetune to SETRSETR ViT-Base · ADE20KGFLOPs: 107.7 -> 80.6Improved
apply DToP@Finetune to SegViT ViT-LargeSegViT · ADE20KmIoU (%): 53.3 -> 52.4 with 617.0 -> 380.3 GFLOPsMixed
apply DToP to Pascal ContextSETR ViT-Base · Pascal ContextmIoU (%): 58.1 -> 58.2Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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