AI Paper Atlas

Research map / Detection and segmentation / CNN features

SemIE: Semantically-Aware Image Extrapolation

Bholeshwar Khurana, Soumya Ranjan Dash, Abhishek Bhatia, Aniruddha Mahapatra, Hrituraj Singh, Kuldeep Kulkarni

ICCV 2021 · Detection and segmentation › CNN features · approach: Unified panoptic segmentation · 13 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 instance-aware context normalization (IaCN) (ablation)SemIE without patch discriminator · CityscapesFID: 48.77 -> 47.76Improved
add patch co-occurrence discriminator (ablation)SemIE without IaCN · CityscapesFID: 48.77 -> 48.72No clear effect
add patch co-occurrence discriminator to model with IaCN (ablation)SemIE with IaCN · CityscapesFID: 47.76 -> 47.67No clear effect
add IaCN to model with patch discriminator (ablation)SemIE with D_patch · CityscapesFID: 48.72 -> 47.67Improved
extrapolate in semantic label space with panoptic maps instead of the SPGNet++ pipelineSPGNet++ · CityscapesFID: 52.14 -> 47.67Improved
extrapolate in semantic label space with panoptic maps instead of the SPGNet++ pipelineSPGNet++ · ADE20K-bedroomFID: 97.23 -> 90.45Improved
use a strong SPADE-based generator with multiscale discriminator for SPGNetSPGNet · CityscapesFID: 83.84 -> 52.14Improved
use semantic-label-space extrapolation instead of RGB-space outpainting (Outpainting-SRN)Outpainting-SRN · CityscapesFID: 66.89 -> 47.67Improved
add instance boundary channel supervision in semantic label extrapolation (ablation)POGNet · Cityscapes and ADE20K-bedroomshape realism (qualitative)Improved
use single-stage SPADE extrapolation without label-space stages (ablation)SemIE stage 4 only · Cityscapesnew-object generation (qualitative)Hurt

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