Research map / Detection and segmentation / CNN features
Multi-Exit Semantic Segmentation Networks
Alexandros Kouris, Stylianos I. Venieris, Stefanos Laskaridis, Nicholas Lane
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 exit-aware (exit-dropout) pre-training and frozen-backbone stage vs Frozen (ablation) | MESS DRN-50 uniform exits · MS COCO val | accuracy across exits: up to 12.57pp | Improved |
| use two-stage training instead of end-to-end joint training (ablation) | MESS DRN-50 uniform exits · MS COCO val | accuracy across exits: up to 3.38 pp | Improved |
| train exits end-to-end jointly (E2E) rather than vanilla single-exit (ablation) | DRN-50 backbone · MS COCO val | final-exit accuracy: 1.57 pp accuracy drop in the final exit | Hurt |
| add Positive Filtering Distillation instead of CE (ablation) | MESS second stage · MS COCO val | accuracy: up to 1.8 pp | Improved |
| add Positive Filtering Distillation instead of KD (ablation) | MESS second stage · MS COCO val | accuracy: up to 2.32 pp | Improved |
| add Positive Filtering Distillation instead of SelfDistill (ablation) | MESS second stage · MS COCO val | accuracy: up to 1.28 pp | Improved |
| use image-level confidence metric instead of averaging per-pixel confidence (ablation) | MESS exit policy · MS COCO val | accuracy at equal speed: up to 6.34 pp (1.17 pp on average) | Improved |
| use tailored searched exit architectures instead of uniform FCN heads (ablation) | MESS · MS COCO val | mIoU under latency budget: up to 19.3 pp (10.9 pp on avg.) | Improved |
| use tailored searched exit architectures instead of uniform FCN heads (ablation) | MESS · MS COCO val | FLOPs under accuracy target: up to 3x (2.4x on avg.) | Improved |
| apply MESS to DRN-50 for latency | DRN-50 · MS COCO val | workload: 3.36x workload reduction, 2.23x latency speedup, no accuracy drop | Improved |
| allow <=1 pp accuracy degradation in MESS search (ablation) | MESS DRN-50 · MS COCO val | workload: 4.01x workload, 2.65x latency | Improved |
| search for accuracy under DRN workload budget | MESS DRN-50 · MS COCO val | mIoU: +5.33 pp mIoU with 1.22x fewer GFLOPs | Improved |
Closest papers by meaning
- Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells (CVPR 2019 · similarity 0.88)
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs (TPAMI 2017 · similarity 0.88)
- Region-Based Semantic Segmentation with End-to-End Training (ECCV 2016 · similarity 0.88)
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2025 · similarity 0.86)
- MaSS13K: A Matting-level Semantic Segmentation Benchmark (CVPR 2025 · similarity 0.86)
- ALGM: Adaptive Local-then-Global Token Merging for Efficient Semantic Segmentation with Plain Vision Transformers (CVPR 2024 · similarity 0.86)
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.
