Research map / Detection and segmentation / CNN features
FADE: Fusing the Assets of Decoder and Encoder for Task-Agnostic Upsampling
Hao Lu, Wenze Liu, Hongtao Fu, Zhiguo Cao
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add encoder features to decoder-only kernel generation (ablation) | CARAFE-style upsampling · ADE20K val | mIoU: 42.82 -> 43.27 | Mixed |
| add encoder features to decoder-only kernel generation (ablation) | upsampling kernel generator · Adobe Composition-1K matting | SAD: 41.01 -> 32.41 | Improved |
| add decoder features to encoder-only kernel generation (ablation) | upsampling kernel generator · Adobe Composition-1K matting | SAD: 34.22 -> 32.41 | Improved |
| add decoder features to encoder-only kernel generation (ablation) | upsampling kernel generator · ADE20K val | mIoU: 42.75 -> 43.27 | Improved |
| replace naive concat+conv with semi-shift convolution (ablation) | FADE kernel generation · Adobe Composition-1K matting | SAD: 32.41 -> 31.78 | Improved |
| replace naive concat+conv with semi-shift convolution (ablation) | FADE kernel generation · ADE20K val | bIoU: 31.55 -> 32.06 | Improved |
| add gating mechanism instead of skipping encoder features (ablation) | FADE (semi-shift) · Adobe Composition-1K matting | SAD: 32.64 -> 31.10 | Improved |
| add gating mechanism instead of skipping encoder features (ablation) | FADE (semi-shift) · ADE20K val | mIoU: 43.22 -> 44.41 | Improved |
| add gating mechanism to semi-shift FADE (ablation) | FADE · ADE20K val | mIoU: 43.33 -> 44.41 | Improved |
| add skipping of encoder features to semi-shift FADE (ablation) | FADE · Adobe Composition-1K matting | SAD: 31.78 -> 32.64 | Hurt |
| generate kernels from encoder and decoder features instead of decoder only (ablation) | CARAFE · SUN RGBD segmentation | bIoU: 25.61 -> 28.80 | Improved |
| generate kernels from encoder and decoder features instead of decoder only (ablation) | CARAFE · Fashion-MNIST reconstruction | PSNR: 24.35 -> 33.83 | Improved |
Later papers that compared with this method
2 later accepted papers report FADE as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- FeatUp: A Model-Agnostic Framework for Features at Any Resolution (ICLR 2024 · better on 1 of 2 shared results)
- Learning to Upsample by Learning to Sample (ICCV 2023 · better on 3 of 4 shared results)
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.
- CARAFE — CARAFE: Content-Aware ReAssembly of FEatures (ICCV 2019) · better on 4 of 4 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
- FADE: A Task-Agnostic Upsampling Operator for Encoder–Decoder Architectures (IJCV 2024 · similarity 0.98)
- UPLiFT: Efficient Pixel-Dense Feature Upsampling with Local Attenders (CVPR 2026 · similarity 0.89)
- Upsample Anything: A Simple and Hard to Beat Baseline for Feature Upsampling (CVPR 2026 · similarity 0.88)
- Spectral Prompt Tuning: Unveiling Unseen Classes for Zero-Shot Semantic Segmentation (AAAI 2024 · similarity 0.85)
- Visual Prompting for Generalized Few-shot Segmentation: A Multi-scale Approach (CVPR 2024 · similarity 0.85)
- Rethinking Decoders for Transformer-based Semantic Segmentation: A Compression Perspective (NeurIPS 2024 · similarity 0.84)
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.
