AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

DPF: Learning Dense Prediction Fields With Weak Supervision

Xiaoxue Chen, Yuhang Zheng, Yupeng Zheng, Qiang Zhou, Hao Zhao, Guyue Zhou, Ya-Qin Zhang

CVPR 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 3 citations (OpenAlex)

Read the paper

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 DPF to the DPT baselineDPT · PASCALContextmIoU: 40.4 -> 45.3Improved
add DPF to the DPT baselineDPT · ADE20KmIoU: 30.4 -> 33.8Improved
add DPF to the DPT baselineDPT · IIWWHDR: 14.0 -> 11.9Improved
add auxiliary supervision to the dense prediction backbone (ablation)DPF · PASCALmIoU: 40.0 (-5.3) -> 45.3Improved
add auxiliary supervision to the dense prediction backbone (ablation)DPF · ADE20KmIoU: 28.3 (-5.5) -> 33.8Improved
add auxiliary supervision to the dense prediction backbone (ablation)DPF · IIWWHDR: 25.2 (-13.3) -> 11.9Improved
add the guidance encoder and guidance features (ablation)DPF · PASCALmIoU: 44.5 (-0.8) -> 45.3Improved
add the guidance encoder and guidance features (ablation)DPF · ADE20KmIoU: 32.9 (-0.9) -> 33.8Improved
add the guidance encoder and guidance features (ablation)DPF · IIWWHDR: 12.5 (-0.6) -> 11.9Improved
use learned interpolation weights instead of relative-distance weights (ablation)DPF · PASCALmIoU: 44.2 -> 45.3Improved
use learned interpolation weights instead of relative-distance weights (ablation)DPF · ADE20KmIoU: 32.1 -> 33.8Improved
use learned interpolation weights instead of relative-distance weights (ablation)DPF · IIWWHDR: 12.9 -> 11.9Improved

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