Research map / Architectures and efficiency / RNNs and normalization
MiniSeg: An Extremely Minimum Network for Efficient COVID-19 Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace single-branch DSConv modules with spatial-pyramid multi-branch modules (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 75.78 -> 76.31 | Helped |
| Add attentive hierarchical fusion to the multi-branch modules (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 76.31 -> 76.58 | Helped |
| Add the encoder two-path design (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 76.58 -> 76.66 | Helped |
| Add channel split to the two-path design (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 76.66 -> 78.33 | Helped |
| Use PReLU instead of ReLU (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 76.92 -> 78.33 | Helped |
| Keep the decoder sub-network rather than remove it (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 73.11 -> 78.33 | Helped |
| Keep deep supervision during training (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 76.45 -> 78.33 | Helped |
| Use Convolution Blocks rather than AHSP modules in the first stage (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 76.71 -> 78.33 | Helped |
| Use 5×5 rather than 3×3 DSConv in Downsampler Blocks (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 76.54 -> 78.33 | Helped |
| Use Feature Fusion Modules rather than AHSP modules in the decoder (ablation) | MiniSeg · COVID-19-P1110 | mIoU: 77.15 -> 78.33 | Helped |
Closest papers by meaning
- Automated Model Design and Benchmarking of Deep Learning Models for COVID-19 Detection with Chest CT Scans (AAAI 2021 · similarity 0.82)
- Toward Robust Diagnosis: A Contour Attention Preserving Adversarial Defense for COVID-19 Detection (AAAI 2023 · similarity 0.81)
- VQ-Seg: Vector-Quantized Token Perturbation for Semi-Supervised Medical Image Segmentation (NeurIPS 2025 · similarity 0.79)
- SegVol: Universal and Interactive Volumetric Medical Image Segmentation (NeurIPS 2024 · similarity 0.78)
- Enhancing SAM with Efficient Prompting and Preference Optimization for Semi-supervised Medical Image Segmentation (CVPR 2025 · similarity 0.78)
- LangDAug: Langevin Data Augmentation for Multi-Source Domain Generalization in Medical Image Segmentation (ICML 2025 · similarity 0.78)
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.