AI Paper Atlas

Research map / Detection and segmentation / CNN features

Learning Semantic Neural Tree for Human Parsing

Ruyi Ji, Dawei Du, Libo Zhang, Longyin Wen, Yanjun Wu, Chen Zhao, Feiyue Huang, Siwei Lyu

ECCV 2020 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 67 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 the prediction module for the final parsing output (ablation)SNT · LIP valmIoU: 50.02 -> 54.73Improved
use deformable convolutions in the prediction module (ablation)SNT · LIP valmIoU: 54.42 -> 54.73Improved
add skip connections from backbone residual blocks to semantic aggregation (ablation)SNT · LIP valmIoU: 53.32 -> 54.73Improved
add attention mask in attention routing module (ablation)SNT w/o skip · LIP valmIoU: 52.15 -> 53.32Improved
increase tree height from 0 to 3 (ablation)SNT · LIP valmIoU: 46.34 -> 54.73Improved
increase tree height from 3 to 4 (ablation)SNT · LIP valmIoU: 54.73 -> 51.42Hurt
increase tree height from 2 to 3 (ablation)SNT · LIP valmIoU: 53.32 -> 54.73Improved
use SNT instead of CE2P (w flip)human parsing · LIP valmIoU: 53.10 -> 54.73Improved
use SNT instead of DPChuman parsing · Pascal-Person-Part valmIoU: 71.34 -> 71.59Improved
use SNT instead of M-CE2P at strict IoU 0.7multiple human parsing · CIHP valAP^r_0.7: 29.74 -> 33.00Improved
use SNT instead of M-CE2Pmultiple human parsing · MHP-v2 valPCP_0.5: 43.77 -> 43.50No clear effect

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