AI Paper Atlas

Research map / Image generation / GANs

Shadow Detection With Conditional Generative Adversarial Networks

Vu Nguyen, Tomas F. Yago Vicente, Maozheng Zhao, Minh Hoai, Dimitris Samaras

ICCV 2017 · Image generation › GANs · approach: Stabilized GAN training · 209 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 sensitivity parameter and weighted data loss to cGAN (scGAN vs cGAN) (ablation)cGAN · SBU testBER: 13.6 -> 9.1Improved
add sensitivity parameter and weighted data loss to cGAN (ablation)cGAN · SBU testshadow error rate: 20.5 -> 7.8Improved
add sensitivity parameter and weighted data loss to cGAN (ablation)cGAN · SBU testnon-shadow error rate: 6.9 -> 10.4Hurt
raise test-time sensitivity w from 0.3 to 0.9 (ablation)scGAN · SBU testshadow error rate: 15.4 -> 4.7Improved
raise test-time sensitivity w from 0.3 to 0.9 (ablation)scGAN · SBU testnon-shadow error rate: 5.8 -> 16.5Hurt
choose w = 0.7 instead of 0.3 (ablation)scGAN · SBU testBER: 10.6 -> 9.0Improved
choose w = 0.9 instead of 0.7 (ablation)scGAN · SBU testBER: 9.0 -> 10.6Hurt
use scGAN instead of stackedCNNStackedCNN · SBU testBER: 11.0 -> 9.1Improved
use scGAN instead of stackedCNN, both trained on UCFStackedCNN · UCF testBER: 11.6 -> 10.9Improved
use scGAN instead of stackedCNN, both trained on SBUStackedCNN · UCF test (cross-dataset)BER: 13.0 -> 11.5Improved

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