card: 14-col schema + depth tar.gz docs

README.md

@@ -25,12 +25,12 @@ configs:
 
 A 36,000-image dataset for **cable / wire instance segmentation**: physically simulated
 deformable linear objects (DLOs) rendered over diverse backgrounds, with exact per-instance
-masks. Single category: `cable`. Generated with **DeformX**.
+masks and per-image metric depth. Single category: `cable`. Generated with **DeformX**.
 
-- **Images:** 36,000 (1024×1024 RGB)
+- **Images:** 36,000 (1024x1024 RGB)
 - **Instances:** 176,387 cable masks (COCO RLE)
 - **Categories:** 1 (`cable`)
-- **Modalities:** RGB, instance masks · per-image metric **depth** (forthcoming, see below)
+- **Modalities:** RGB, instance masks, per-image metric depth
 
 ## Scenes
 
@@ -40,15 +40,15 @@ masks. Single category: `cable`. Generated with **DeformX**.
 | `flying_wire`   | 2 / 4 / 8  | 15,000 | suspended / catenary wires |
 | `wire_on_plane` | 2 / 4 / 8  | 15,000 | wires draped on a flat surface |
 
-Each `scene_family/<n>_cable` group is uniformly sampled; every image has ≥0.5% wire pixels.
-
 ## Files & layout
 
 | path | what |
 |---|---|
 | `data/*.parquet` (config `default`) | RGB image + per-instance COCO-RLE masks + metadata, one row per image |
 | `annotations.json` | full COCO ground truth (36,000 images / 176,387 annotations, single category `cable`) |
-| `depth/…` | per-image metric depth, float32 `[1024, 1024]` — **forthcoming** |
+| `depth/*.tar.gz` | per-image metric depth, float32 `[1024,1024]` `.npy`, packed in 100 gzipped tar shards |
+
+> The `image-only` branch holds the RGB+masks release without depth.
 
 ## Usage
 
@@ -64,14 +64,21 @@ masks = json.loads(ex["masks_rle"])      # list of {"size":[h,w], "counts": <RLE
 m0 = M.decode(masks[0])                   # HxW uint8 instance mask
 ```
 
-The COCO `annotations.json` keys images by the same `file_name`.
+**Depth.** Download and extract the `depth/*.tar.gz` shards; each member is
+`<scene_family>/<n>_cable/<id>.npy`. The depth for an image is the `.npy` at the same
+relative path as its `file_name` (swap `.png` -> `.npy`):
+
+```python
+import numpy as np
+d = np.load(f"depth/{ex['file_name'][:-4]}.npy")   # float32 [1024,1024], metric
+```
 
 ## Data fields (Parquet `default`)
 
 `image` (RGB), `file_name`, `scene_family`, `n_cable`, `masks_rle`, `bboxes`, `areas`,
-`num_instances`, `wire_area_frac`, `bg_area_frac`, `det_f1_iou_50`, `det_f1_iou_75`,
-`det_map_50_95`, `jaccard_j`, `category`. Masks / bboxes / areas are JSON-encoded per-instance
-lists (RLE = COCO compressed). The dataset license (CC-BY-4.0) applies to all rows.
+`num_instances`, `wire_area_frac`, `det_f1_iou_50`, `det_f1_iou_75`, `det_map_50_95`,
+`jaccard_j`, `category`. Masks / bboxes / areas are JSON-encoded per-instance lists
+(RLE = COCO compressed).
 
 ## Splits