README: code-only repo; unified-512 resolution-fair protocol (done)

README.md

@@ -5,22 +5,28 @@ tags: [medical-imaging, segmentation, benchmark]
 
 # GenSeg-Baselines
 
-Baseline benchmark for 2D medical image segmentation: **8 methods x 10 datasets x 3 seeds/folds, 7 metrics**.
-Companion to the [GenSegDataset](https://huggingface.co/datasets/MaybeRichard/GenSegDataset).
+Reproducible **code** for a 2D medical-image segmentation benchmark: **8 methods × 10 datasets × 3 seeds/folds, 7 metrics**, evaluated under a **unified resolution-fair protocol**. Companion to the [GenSegDataset](https://huggingface.co/datasets/MaybeRichard/GenSegDataset).
 
-**Methods:** UNet, UNet++, DeepLabV3+ (ResNet-50/ImageNet), Attention-UNet (scratch),
-TransUNet (R50-ViT-B/16), Swin-UNet (Swin-Tiny), nnU-Net v2 (250ep), U-Mamba (UMambaBot, 100ep).
+This is a **code-only** repository — trained checkpoints and the generated result tables are not hosted here.
+
+**Methods:** UNet, UNet++, DeepLabV3+ (ResNet-50/ImageNet), Attention-UNet (from scratch),
+TransUNet (R50-ViT-B/16, input 256), Swin-UNet (Swin-Tiny, input 224), nnU-Net v2 (250 ep), U-Mamba (UMambaBot, 100 ep).
 
 **Datasets:** cvc_clinicdb, kvasir_seg, fives, busi, refuge2, acdc, idridd, pannuke, isic2018, kits19.
 
-**Metrics:** Dice, IoU, HD95, ASSD, Sensitivity, Specificity, Precision (+ efficiency).
+**Metrics (computed per image, then aggregated):** Dice, IoU, HD95, ASSD, Sensitivity, Specificity, Precision —
+plus per-class Dice for the multi-class datasets and paired-Wilcoxon significance on per-image Dice.
 
-## Layout
-- `code/` - baseline framework (train/test/aggregate), scripts, conda envs. *(Generative SegGen code excluded.)*
-- `results/` - per-run `metrics.json` + aggregated `summary.{html,csv,md,tex}` + `efficiency.md`.
-- `weights/` - curated checkpoints: best seed per (dataset, arch) for framework; best fold for nnU-Net / U-Mamba.
+## Resolution-fair protocol
+Convolutional nets are trained at 512; the fixed-input transformers (Swin-UNet 224, TransUNet 256) and
+nnU-Net / U-Mamba run at their native size; **every prediction and ground truth is resized to a common
+512×512 before scoring**, so boundary metrics (HD95/ASSD, in pixels) are directly comparable across methods.
 
-## Note
-These are the **256-px baseline** (confirmed). A resolution-fair re-evaluation (conv methods retrained at a
-higher per-dataset resolution; all methods scored at a common R so HD95 is comparable) is in progress and
-will be added later.
+## Layout (code only)
+- `code/framework/` — training/evaluation framework: `train.py`, `test.py`, `eval_at_res.py`,
+  `nnunet_eval.py`; `metrics/` (the 7 metrics + boundary distances); `models/` (SMP wrappers,
+  Attention-UNet, Swin/TransUNet wrappers, model registry); `report/aggregate.py` builds the summary
+  tables (per-dataset Dice/HD95/IoU, per-class Dice, Sensitivity/Precision, significance).
+- `code/sota/{Swin-Unet,TransUNet}/` — upstream network definitions imported by the Swin-UNet / TransUNet wrappers.
+- `code/scripts/` — reproduction scripts (unified-512 training & evaluation, nnU-Net / U-Mamba pipelines).
+- `code/envs/` — conda environments (`seggen.yml`, `nnunet.yml`, `umamba.yml`).