Add README with Image feature schema

README.md

@@ -0,0 +1,93 @@
+---
+license: etalab-2.0
+language:
+- fr
+tags:
+- panoramax
+- street-view
+- france
+- gps
+- ign
+size_categories:
+- 1M<n<10M
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/*.parquet
+dataset_info:
+  features:
+  - name: id
+    dtype: string
+  - name: image
+    dtype: image
+  - name: collection_id
+    dtype: string
+  - name: lat
+    dtype: float64
+  - name: lng
+    dtype: float64
+  - name: azimuth
+    dtype: float64
+  - name: datetime
+    dtype: timestamp[us, tz=UTC]
+  - name: sequence_rank
+    dtype: int32
+  - name: width
+    dtype: int32
+  - name: height
+    dtype: int32
+  - name: panoramax_url
+    dtype: string
+  - name: producer
+    dtype: string
+  - name: license
+    dtype: string
+  - name: original_filename
+    dtype: string
+---
+
+# Panoramax sig14 photos
+
+Street-view photos from the IGN Panoramax instance, scraped from the user **sig14**
+(Service d'Information Géographique du Calvados, France).
+
+Each row is one street-level photo plus its GPS coordinates and camera heading.
+
+## Source
+
+- API: <https://panoramax.ign.fr/api>
+- User: `sig14` (UUID `5215eca1-4a1b-40eb-a96c-5dc7ff2cc6d8`)
+- License: [Etalab 2.0 (Open License)](https://alliance.numerique.gouv.fr/licence-ouverte-open-licence/)
+
+## Schema
+
+| field | type | description |
+|---|---|---|
+| `id` | string | Panoramax item UUID |
+| `image` | Image | JPEG bytes, HD quality |
+| `collection_id` | string | Sequence (collection) UUID |
+| `lat`, `lng` | float64 | WGS84 coordinates |
+| `azimuth` | float64 | Camera heading, 0–360° (0 = North, 90 = East) |
+| `datetime` | timestamp(UTC) | When the photo was taken |
+| `sequence_rank` | int32 | Position of this photo in its sequence |
+| `width`, `height` | int32 | Image dimensions in pixels |
+| `panoramax_url` | string | Link back to the photo on panoramax.ign.fr |
+| `producer` | string | Always `sig14` for this dataset |
+| `original_filename` | string | Original filename on IGN side |
+
+## Storage layout
+
+One Parquet shard per Panoramax collection (sequence) — file naming
+`data/collection={uuid}.parquet`. Idempotent updates: re-running the scraper
+only fetches collections not yet present in the dataset.
+
+## Loading
+
+```python
+from datasets import load_dataset
+ds = load_dataset("rtrm/panoramax-sig14-photos", split="train", streaming=True)
+for row in ds.take(5):
+    print(row["lat"], row["lng"], row["azimuth"])
+    row["image"].save(f'{row["id"]}.jpg')
+```