--- a
+++ b/README.md
@@ -0,0 +1,53 @@
+---
+task_categories:
+- image-classification
+
+---
+# AutoTrain Dataset for project: stroke-classifier
+
+## Dataset Description
+
+This dataset has been automatically processed by AutoTrain for project stroke-classifier.
+
+### Languages
+
+The BCP-47 code for the dataset's language is unk.
+
+## Dataset Structure
+
+### Data Instances
+
+A sample from this dataset looks as follows:
+
+```json
+[
+  {
+    "image": "<233x197 L PIL image>",
+    "target": 0
+  },
+  {
+    "image": "<233x197 L PIL image>",
+    "target": 0
+  }
+]
+```
+
+### Dataset Fields
+
+The dataset has the following fields (also called "features"):
+
+```json
+{
+  "image": "Image(decode=True, id=None)",
+  "target": "ClassLabel(names=['notStroke', 'stroke'], id=None)"
+}
+```
+
+### Dataset Splits
+
+This dataset is split into a train and validation split. The split sizes are as follow:
+
+| Split name   | Num samples         |
+| ------------ | ------------------- |
+| train        | 1600 |
+| valid        | 945 |