Research map / Detection and segmentation / Open-set and few-shot detection
FGNet: Leveraging Feature-Guided Attention to Refine SAM2 for 3D EM Neuron Segmentation
Zhenghua Li, Hang Chen, Zihao Sun, Kai Li, Xiaolin Hu
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 SAM2 encoder (ablation) | FGNet · AC3/AC4 | VOI: 1.108 -> 0.797 | Improved |
| add Fine-Grained Encoder (ablation) | FGNet · AC3/AC4 | VOI: 1.355 -> 0.797 | Improved |
| add Feature-Guided Attention between SAM2 and FGE (ablation) | FGNet · AC3/AC4 | VOI: 1.051 -> 0.797 | Improved |
| fine-tune SAM2 weights rather than keep them frozen (ablation) | FGNet · AC3/AC4 | VOI: 0.910 -> 0.797 | Improved |
| fine-tune SAM2 weights rather than keep them frozen (ablation) | FGNet · CREMI-B | VOI: 0.778 -> 0.677 | Improved |
| use FGNet with frozen SAM2 instead of CAD | AC3/AC4 baseline · AC3/AC4 | VOI: 1.032 -> 0.910 | Mixed |
| use FGA module instead of SE-type attention (ablation) | FGNet · AC3/AC4 | VOI: 0.979 -> 0.797 | Improved |
| use FGA module instead of CBAM-type attention (ablation) | FGNet · AC3/AC4 | VOI: 0.883 -> 0.797 | Improved |
| use FGA module instead of ECA-type attention (ablation) | FGNet · AC3/AC4 | VOI: 0.839 -> 0.797 | Improved |
| increase FGE depth from 4 to 5 (ablation) | FGNet · AC3/AC4 | VOI: 0.797 -> 0.883 | No clear effect |
| decrease FGE depth from 4 to 3 (ablation) | FGNet · AC3/AC4 | VOI: 0.797 -> 0.816 | No clear effect |
| use FGA-guided refinement instead of full fine-tuning of SAM2 | SAM2 adaptation · AC3/AC4 | VOI: 1.355 -> 0.797 | Improved |
Closest papers by meaning
- Cross-Dimension Affinity Distillation for 3D EM Neuron Segmentation (CVPR 2024 · similarity 0.90)
- Learning Multimodal Volumetric Features for Large-Scale Neuron Tracing (AAAI 2024 · similarity 0.87)
- $\epsilon$-Seg: Sparsely Supervised Semantic Segmentation of Microscopy Data (NeurIPS 2025 · similarity 0.86)
- Electron Microscopy Images as Set of Fragments for Mitochondrial Segmentation (AAAI 2024 · similarity 0.86)
- Learning Large-Factor EM Image Super-Resolution with Generative Priors (CVPR 2024 · similarity 0.85)
- FISBe: A Real-World Benchmark Dataset for Instance Segmentation of Long-Range Thin Filamentous Structures (CVPR 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.
