AI Paper Atlas

Research map / Image generation / Editing and style transfer

Cross-Domain Cascaded Deep Translation

Oren Katzir, Dani Lischinski, Daniel Cohen-Or

ECCV 2020 · Image generation › Editing and style transfer · approach: Image-to-image translation

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
use cascaded deep-to-shallow translation instead of Layer 5 only (ablation)Cascaded translation · Cat<->DogFID: 126.93/127.53 -> 67.58/46.02Improved
use cascaded translation instead of Layer 4 only (ablation)Cascaded translation · Zebra<->GiraffeFID: 103.41/53.36 -> 67.41/39.38Improved
use cascaded translation instead of Layer 3 only (ablation)Cascaded translation · Zebra<->ElephantFID: 166.32/113.28 -> 68.45/47.86Improved
use cascaded translation instead of Layer 5 only (ablation)Cascaded translation · Zebra<->ElephantFID: 101.26/76 -> 68.45/47.86Improved
use cascaded translation instead of Layer 5 only (ablation)Cascaded translation · Zebra<->GiraffeFID: 167.62/184.37 -> 67.41/39.38Improved
translate shallower VGG layers only instead of deep layers (ablation)Single-layer translation · shape deformation (qualitative)visual qualityHurt
add cycle, identity and adversarial losses together (ablation)Deepest layer translator · layer 5 translationvisual qualityImproved
fine-tune VGG-19 on zebra vs giraffe classification (ablation)Cascaded translation · Zebra<->Giraffevisual qualityHurt
use AlexNet features instead of VGG-19 (ablation)Cascaded translation · Zebra<->Giraffevisual qualityHurt
use WGAN-GP instead of LSGAN adversarial lossDeepest layer translator · deep feature translationtraining stabilityImproved
translate deep features rather than images (vs GANimorph)Cascaded translation · Zebra<->GiraffeFID: 98.25/120.05 -> 67.41/39.38Mixed
synthesize VGG deep features then invert (DEEP-VAE) instead of VQ-VAE-2DEEP-VAE · Zebra synthesisFID: 154.41 -> 57.66Improved

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