Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,52 @@
+---
+license: cc-by-4.0
+task_categories:
+- image-segmentation
+- object-detection
+tags:
+- recycling
+- steel-scrap
+- copper
+- yolo
+configs:
+- config_name: a1
+  data_files:
+  - split: test
+    path: data/a1/test/*.parquet
+  - split: train
+    path: data/a1/train/*.parquet
+  - split: val
+    path: data/a1/val/*.parquet
+---
+# SteelDS (Parquet mirror)
+
+Shredded E40 steel scrap and copper on a lab conveyor belt, 3680 x 2160 frames with YOLO segmentation labels.
+Mirror of https://zenodo.org/records/20271102, converted for streaming (videos left out). Licence: CC BY 4.0, credit the original authors.
+
+- Subsets (configs): a1; a1-a3 are labelled, a4-a5 unlabelled
+- Classes: 0 = background, 1 = steel, 2 = copper
+- `yolo_label`: the original YOLO-seg label text ("<class> x1 y1 x2 y2 ..." normalised to 0-1); `has_label` = False where the original frame had no label file
+
+## Load
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("vivekkol/steelds", "a1", split="train", streaming=True)   # or without streaming=True to download once
+for row in ds:
+    image = row["image"]          # PIL image
+    label = row["yolo_label"]     # YOLO-seg text, same as the original .txt file
+```
+
+## Write out as YOLO folders (for Ultralytics)
+
+```python
+import os
+from datasets import load_dataset
+
+for split in ["train", "val", "test"]:
+    os.makedirs(f"steelds/images/{split}", exist_ok=True); os.makedirs(f"steelds/labels/{split}", exist_ok=True)
+    for row in load_dataset("vivekkol/steelds", "a1", split=split, streaming=True):
+        row["image"].save(f"steelds/images/{split}/{row['file_name']}.jpg", quality=95)
+        open(f"steelds/labels/{split}/{row['file_name']}.txt", "w").write(row["yolo_label"])
+```