AI Paper Atlas

Research map / Detection and segmentation / CNN features

Illuminating Pedestrians via Simultaneous Detection & Segmentation

Garrick Brazil, Xi Yin, Xiaoming Liu

ICCV 2017 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation

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 weak segmentation supervision (segmentation infusion layer) (ablation)SDS-RCNN · Caltech reasonablefused miss rate (%): 10.41 -> 7.36Improved
add weak segmentation supervision (ablation)SDS-RCNN RPN · Caltech reasonableRPN miss rate (%): 13.84 -> 10.67Improved
add weak segmentation supervision (ablation)SDS-RCNN BCN · Caltech reasonableBCN miss rate (%): 18.76 -> 10.98Improved
add 20% proposal padding (ablation)SDS-RCNN · Caltech reasonablefused miss rate (%): 7.69 -> 7.36Improved
add 20% proposal padding (ablation)SDS-RCNN BCN · Caltech reasonableBCN miss rate (%): 13.09 -> 10.98Improved
add cost-sensitive weighting (ablation)SDS-RCNN RPN · Caltech reasonableRPN miss rate (%): 9.63 -> 10.67Hurt
add cost-sensitive weighting (ablation)SDS-RCNN BCN · Caltech reasonableBCN miss rate (%): 14.87 -> 10.98Improved
add cost-sensitive weighting (ablation)SDS-RCNN · Caltech reasonablefused miss rate (%): 7.89 -> 7.36Improved
use strict labeling (IoU>0.7) instead of standard (IoU>0.5) for BCN (ablation)SDS-RCNN · Caltech reasonablefused miss rate (%): 8.71 -> 7.36Improved
use strict labeling for BCN (ablation)SDS-RCNN BCN · Caltech reasonableBCN miss rate (%): 17.41 -> 10.98Improved
share conv3 features between RPN and BCN instead of separate networks (ablation)SDS-RCNN · Caltech reasonablefused miss rate (%): 7.36 -> 8.66Hurt
share conv5 features between RPN and BCN (ablation)SDS-RCNN · Caltech reasonablefused miss rate (%): 7.36 -> 10.87Hurt

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