AI Paper Atlas

Research map / Image generation / Image restoration

CompositeTasking: Understanding Images by Spatial Composition of Tasks

Nikola Popovic, Danda Pani Paudel, Thomas Probst, Guolei Sun, Luc Van Gool

CVPR 2021 · Image generation › Image restoration · approach: Dehazing and deraining · 5 citations (OpenAlex)

Check 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 spatial task conditioning with one shared encoder-decoder, trained on random mosaic label regionsCTN · PASCAL-MT, single-task rule SmIoU: 59.82 -> 62.45Improved
Use spatial task conditioning with one shared encoder-decoder, trained on random mosaic label regionsCTN · PASCAL-MT, single-task rule SmErr: 16.07 -> 16.93Hurt
Use spatial task conditioning for sparse, semantically assigned task regionsCTN · PASCAL-MT, semantic rule R2odsF: 63.90 -> 69.20Improved
Use spatial task conditioning for sparse, semantically assigned task regionsCTN · PASCAL-MT, semantic rule R2mErr: 17.13 -> 18.12Hurt
Fine-tune the R2-trained model on the similar R3 ruleCTN · PASCAL-MT, semantic rule R3mErr: 20.11 -> 18.68Improved
Fine-tune the R2-trained model on R3CTN · PASCAL-MT, semantic rule R2mErr: 18.12 -> 17.95Improved
Train on independently random task assignments instead of large random mosaicsCTN · PASCAL-MT, random task per pixel RrndmIoU: 35.89 -> 59.58Improved
Train on independently random task assignments instead of large random mosaicsCTN · PASCAL-MT, random task per pixel RrndmErr: 64.36 -> 17.16Improved
Use 2 shared embedding convolutions instead of 1 (ablation)CTN · PASCAL-MT, single-task rule SmIoU: 62.45 -> 61.93Hurt
Replace 3 × 3 decoder convolution with 1 × 1 (ablation)CTN · PASCAL-MT, single-task rule SmIoU: 62.45 -> 61.50Hurt
Use 6 output channels instead of 3 (ablation)CTN · PASCAL-MT, single-task rule SmIoU: 62.45 -> 62.35No clear effect
Replace bilinear with nearest-neighbor palette-embedding interpolation (ablation)CTN · PASCAL-MT, single-task rule SmIoU: 62.45 -> 60.92Hurt

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