Papers › MSDNet: Multi-Scale Decoder for Few-Shot Semantic Segmentation via Transformer-Guided...
MSDNet: Multi-Scale Decoder for Few-Shot Semantic Segmentation via Transformer-Guided Prototyping
Amirreza Fateh, Mohammad Reza Mohammadi, Mohammad Reza Jahed Motlagh
Few-shot Semantic Segmentation addresses the challenge of segmenting objects in query images with only a handful of annotated examples. However, many previous state-of-the-art methods either have to discard intricate local semantic features or suffer from high computational complexity. To address these challenges, we propose a new Few-shot Semantic Segmentation framework based on the transformer architecture. Our approach introduces the spatial transformer decoder and the contextual mask generation module to improve the relational understanding between support and query images. Moreover, we introduce a multi-scale decoder to refine the segmentation mask by incorporating features from different resolutions in a hierarchical manner. Additionally, our approach integrates global features from intermediate encoder stages to improve contextual understanding, while maintaining a lightweight structure to reduce complexity. This balance between performance and efficiency enables our method to achieve state-of-the-art results on benchmark datasets such as PASCAL-5ⁱ and COCO-20ⁱ in both 1-shot and 5-shot settings. Notably, our model with only 1.5 million parameters demonstrates competitive performance while overcoming limitations of existing methodologies. https://github.com/amirrezafateh/MSDNet
Code
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Tasks
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Few-Shot Semantic Segmentation | COCO-20i (1-shot) | MSDNet (ResNet-101) | FB-IoU | 71.3 | #14 of 85 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (1-shot) | MSDNet (ResNet-101) | Mean IoU | 48.5 | #14 of 85 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (1-shot) | MSDNet (ResNet-101) | learnable parameters (million) | 1.5 | #14 of 85 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (1-shot) | MSDNet (ResNet-50) | FB-IoU | 70.4 | #25 of 85 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (1-shot) | MSDNet (ResNet-50) | Mean IoU | 46.5 | #25 of 85 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (1-shot) | MSDNet (ResNet-50) | learnable parameters (million) | 1.5 | #25 of 85 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (5-shot) | MSDNet (ResNet-101) | FB-IoU | 75.1 | #16 of 81 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (5-shot) | MSDNet (ResNet-101) | Mean IoU | 55.3 | #16 of 81 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (5-shot) | MSDNet (ResNet-101) | learnable parameters (million) | 1.5 | #16 of 81 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (5-shot) | MSDNet (ResNet-50) | FB-IoU | 74.5 | #19 of 81 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (5-shot) | MSDNet (ResNet-50) | Mean IoU | 54.5 | #19 of 81 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i (5-shot) | MSDNet (ResNet-50) | learnable parameters (million) | 1.5 | #19 of 81 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i -> Pascal VOC (1-shot) | MSDNet (ResNet-101) | Mean IoU | 73.9 | #1 of 13 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i -> Pascal VOC (1-shot) | MSDNet (ResNet-50) | Mean IoU | 72.1 | #2 of 13 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i -> Pascal VOC (5-shot) | MSDNet (ResNet-101) | Mean IoU | 76.4 | #5 of 12 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | COCO-20i -> Pascal VOC (5-shot) | MSDNet (ResNet-50) | Mean IoU | 74.2 | #6 of 12 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (1-Shot) | MSDNet (ResNet-101) | FB-IoU | 77.3 | #54 of 105 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (1-Shot) | MSDNet (ResNet-101) | Mean IoU | 64.7 | #54 of 105 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (1-Shot) | MSDNet (ResNet-101) | learnable parameters (million) | 1.5 | #54 of 105 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (1-Shot) | MSDNet (ResNet-50) | FB-IoU | 77.1 | #59 of 105 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (1-Shot) | MSDNet (ResNet-50) | Mean IoU | 64.3 | #59 of 105 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (1-Shot) | MSDNet (ResNet-50) | learnable parameters (million) | 1.5 | #59 of 105 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (5-Shot) | MSDNet (ResNet-101) | FB-IoU | 85.0 | #35 of 96 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (5-Shot) | MSDNet (ResNet-101) | Mean IoU | 70.8 | #35 of 96 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (5-Shot) | MSDNet (ResNet-101) | learnable parameters (million) | 1.5 | #35 of 96 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (5-Shot) | MSDNet (ResNet-50) | FB-IoU | 82.1 | #53 of 96 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (5-Shot) | MSDNet (ResNet-50) | Mean IoU | 68.7 | #53 of 96 | Archive leaderboard | report |
| Few-Shot Semantic Segmentation | PASCAL-5i (5-Shot) | MSDNet (ResNet-50) | learnable parameters (million) | 1.5 | #53 of 96 | Archive leaderboard | report |
Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.
Methods
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections