Research map / Detection and segmentation / CNN features
Learned Watershed: End-To-End Learning of Seeded Segmentation
Steffen Wolf, Lukas Schott, Ullrich Kothe, Fred Hamprecht
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 |
|---|---|---|---|
| use dynamic prediction f_dyn instead of static f_static (ablation) | structured-loss watershed · Artificial data (noise 0.3) | ARAND (%): 6.4 -> 5.8 | Improved |
| use dynamic prediction f_dyn instead of static f_static (ablation) | structured-loss watershed · Artificial data (noise 0.6) | ARAND (%): 13.8 -> 12.5 | Improved |
| use dynamic prediction f_dyn instead of static f_static (ablation) | structured-loss watershed · Artificial data (noise 0.9) | ARAND (%): 32.4 -> 32.2 | No clear effect |
| train with structured loss L_SL and f_static instead of unstructured CNN plus watershed (ablation) | CNN + WS · Artificial data (noise 0.6) | ARAND (%): 14.9 -> 13.8 | Improved |
| use CNN boundary prediction instead of raw image (ablation) | watershed · Artificial data (noise 0.6) | ARAND (%): 41.9 -> 14.9 | Improved |
| swap standard WS for Power WS on CNN output (ablation) | NN + WS · Artificial data (noise 0.6) | ARAND (%): 14.9 -> 14.9 | No clear effect |
| use Learned WS instead of DTWS | DTWS · CREMI neurite segmentation (2D slices) | ARAND: 0.085 -> 0.082 | Improved |
| use Learned WS instead of DTWS | DTWS · CREMI neurite segmentation (2D slices) | VOI merge: 0.070 -> 0.057 | Improved |
| use Learned WS instead of DTWS | DTWS · CREMI neurite segmentation (2D slices) | VOI split: 0.320 -> 0.319 | No clear effect |
| use Learned WS instead of Stochastic WS | Stochastic WS · CREMI neurite segmentation (2D slices) | ARAND: 0.193 -> 0.082 | Improved |
Closest papers by meaning
- Deep Watershed Transform for Instance Segmentation (CVPR 2017 · similarity 0.87)
- The Mutex Watershed: Efficient, Parameter-Free Image Partitioning (ECCV 2018 · similarity 0.86)
- The Mutex Watershed and its Objective: Efficient, Parameter-Free Graph Partitioning (TPAMI 2020 · similarity 0.86)
- Learning on the Edge: Investigating Boundary Filters in CNNs (IJCV 2019 · similarity 0.85)
- Factorized Diffusion Architectures for Unsupervised Image Generation and Segmentation (NeurIPS 2024 · similarity 0.83)
- A Generative Approach at the Instance-Level for Image Segmentation Under Limited Training Data Conditions (Student Abstract) (AAAI 2025 · similarity 0.83)
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.
