Initial Commit

README.md

@@ -1,5 +1,5 @@
 ---
-license: mit
+license: cc-by-nc-4.0
 task_categories:
 - image-classification
 language:
@@ -11,4 +11,31 @@ tags:
 pretty_name: Jaundice Dataset
 size_categories:
 - n<1K
----
+---
+
+# Diverse Jaundice Eye Dataset
+
+## Overview
+This dataset contains web-scraped images of human eyes, categorized into **jaundice** (scleral icterus) and **healthy** eyes. 
+
+A primary goal of this dataset is to mitigate skin tone bias often found in medical imaging. Initial web searches for jaundice disproportionately returned images of lighter skin. To build a more clinically representative dataset for fairer AI training, specific searches (like "jaundice in black people") were used to ensure the inclusion of darker skin tones.
+
+## Dataset Details
+* **Classes:** `jaundice` and `healthy`
+* **Total Images:** 321
+* **Collection Method:** Web scraping
+
+## Folder Structure
+dataset/
+├── train/
+          ├──jaundice/   # Images of eyes showing signs of jaundice
+└── test/
+        ├──healthy/    # Images of healthy eyes
+
+## Limitations & Disclaimer
+Not Clinical Grade: These images are scraped from the web and have not been validated by doctors. This dataset is strictly for educational, prototyping, and research purposes.
+
+Variance: Due to the nature of web scraping, lighting, camera quality, and zoom levels vary across the images.
+
+## Author
+Daniel EKong — [Daniel's Git](https://github.com/Bleachcarte)