Create README_eye-conjectiva_MedThinkSeg.md

eye-conjectiva/README_eye-conjectiva_MedThinkSeg.md

@@ -0,0 +1,107 @@
+# eye-conjectiva (Eye Conjunctiva Segmentation Dataset) — packaged in `saiteja33/MedThink-Seg/eye-conjectiva`
+
+This README documents:
+1) **The original Eye Conjunctiva Segmentation Dataset** (what it contains, size, and original file structure), and  
+2) **How the dataset is organized inside this Hugging Face repo** (`saiteja33/MedThink-Seg`, under the `eye-conjectiva/` directory).
+
+---
+
+## 1) Original dataset (source)
+
+**Original dataset record (Mendeley Data, Version 1):**  
+https://data.mendeley.com/datasets/yxwjgcndg2/1  
+DOI: `10.17632/yxwjgcndg2.1`
+
+### What kind of images are included?
+- **Modality:** RGB photographs of the human **eye conjunctiva**.
+- **Acquisition:** captured using two smartphones (**OnePlus 9R** and **OnePlus 9 Pro**) under adequate room lighting; lower eyelid gently pulled down to expose conjunctiva.
+- **Image size / resolution:** images are **high-resolution ~3000 × 4000** pixels; **masks match the same resolution**.
+- **Annotation tool:** labeled with **Labelme**.
+
+### Dataset size and counts
+- **Total images:** **547** conjunctiva eye images.
+
+### Labels / annotations
+- mask sets are provided:
+  - **Masks Annotator 1**
+
+> The Mendeley record does not specify the exact pixel-value encoding (e.g., binary 0/255). Inspect a few masks to confirm your expected mask format.
+
+
+## 2) Dataset structure in this Hugging Face repo (`saiteja33/MedThink-Seg/eye-conjectiva`)
+
+**HF path:** `eye-conjectiva/`  
+**Repo size (HF UI):** ~**4.58 GB**
+
+### Directory structure
+This packaging contains only images + masks:
+
+```
+eye-conjectiva/
+├─ images/        # conjunctiva images (PNG)
+└─ masks/         # segmentation masks (PNG), same filenames as images/
+```
+
+### Naming convention (HF packaging)
+- Filenames are numeric, like:
+  - `1.png`, `10.png`, `100.png`, ...
+- **Image ↔ mask pairing rule:** `eye-conjectiva/images/<name>.png` pairs with `eye-conjectiva/masks/<name>.png`.
+
+### File sizes (HF packaging)
+- Images are large PNGs (typically several MB each).
+- Masks are much smaller PNGs (typically a few KB to tens of KB).
+
+> Note: The original dataset provides **two** annotators’ masks; this HF packaging exposes a **single** `masks/` folder. If you need both annotators, download from Mendeley.
+
+---
+
+## How to download / use (recommended)
+
+### A) Download only `eye-conjectiva/` from this HF repo
+```python
+from huggingface_hub import snapshot_download
+
+local_dir = snapshot_download(
+    repo_id="saiteja33/MedThink-Seg",
+    repo_type="dataset",
+    allow_patterns=["eye-conjectiva/**"],  # only pull this subdir
+)
+print(local_dir)
+```
+
+### B) Create (image, mask) pairs
+```python
+from pathlib import Path
+
+root = Path(local_dir) / "eye-conjectiva"
+img_dir = root / "images"
+msk_dir = root / "masks"
+
+images = sorted(img_dir.glob("*.png"))
+pairs = [(p, msk_dir / p.name) for p in images]
+
+# Quick checks
+missing = [p.name for p, m in pairs if not m.exists()]
+print("num_images:", len(images))
+print("missing_masks:", len(missing))
+print("example_pair:", pairs[0])
+```
+
+### C) Load with 🤗 Datasets (images only)
+If you want to load just the images as a dataset:
+
+```python
+from datasets import load_dataset
+
+ds_imgs = load_dataset("imagefolder", data_dir=str(img_dir))
+print(ds_imgs)
+```
+
+(For paired image+mask training, most people keep pairs via the folder structure as in section **B**.)
+
+---
+
+## Citation
+If you use this dataset, please cite the Mendeley record:
+
+- *Eye Conjunctiva Segmentation Dataset* (Mendeley Data, V1). DOI: `10.17632/yxwjgcndg2.1`