Research map / Detection and segmentation / CNN features
FADE: A Task-Agnostic Upsampling Operator for Encoder–Decoder Architectures
Hao Lü, 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 (b2) · ADE20K SegFormer-B1 | mIoU: 42.82 -> 43.27 | Improved |
| add decoder features to encoder-only kernel generation (ablation) | b1 · ADE20K SegFormer-B1 | mIoU: 42.75 -> 43.27 | Improved |
| use semi-shift convolution instead of naive kernel generation (ablation) | FADE · ADE20K SegFormer-B1 | mIoU: 43.27 -> 43.33 | Improved |
| use semi-shift convolution instead of naive kernel generation (ablation) | FADE · Composition-1K | SAD: 32.41 -> 31.78 | Improved |
| add gating instead of skipping encoder features (ablation) | FADE · ADE20K SegFormer-B1 | mIoU: 43.22 -> 44.41 | Improved |
| add gating mechanism (ablation) | FADE b4 · ADE20K SegFormer-B1 | mIoU: 43.33 -> 44.41 | Improved |
| add gate to nearest-neighbor upsampling (ablation) | SegNet NN · Weizmann Horse | mIoU: 89.15 -> 95.26 | Improved |
| add gate to CARAFE (ablation) | SegNet CARAFE · Weizmann Horse | mIoU: 89.29 -> 95.25 | Improved |
| use iterative x2 upsampling instead of direct (ablation) | FADE · ADE20K SegFormer-B1 | mIoU: 42.89 -> 44.41 | Improved |
| add FADE to strong SegFormer-B5 backbone | SegFormer-B5 · ADE20K | mIoU: 51.00 -> 50.90 | No clear effect |
| add FADE to SegFormer-B5 backbone, boundary metric | SegFormer-B5 · ADE20K | bIoU: 37.81 -> 38.91 | Improved |
| add gating to FADE for object detection (ablation) | FADE · COCO Faster R-CNN R50 | AP: 38.5 -> 37.8 | 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.
- CARAFE — CARAFE: Content-Aware ReAssembly of FEatures (ICCV 2019) · better on 2 of 3 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: Fusing the Assets of Decoder and Encoder for Task-Agnostic Upsampling (ECCV 2022 · similarity 0.98)
- Visual Prompting for Generalized Few-shot Segmentation: A Multi-scale Approach (CVPR 2024 · similarity 0.85)
- Large Vision-Language Models Get Lost in Attention (ICML 2026 · similarity 0.85)
- An Empirical Study of Attention and Diversity for Adaptive Visual Token Pruning in Large Vision-Language Models (ICLR 2026 · similarity 0.85)
- ProKeR: A Kernel Perspective on Few-Shot Adaptation of Large Vision-Language Models (CVPR 2025 · similarity 0.85)
- Rethinking Decoders for Transformer-based Semantic Segmentation: A Compression Perspective (NeurIPS 2024 · similarity 0.85)
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.
