Update dataset card for v2.0

README.md

@@ -62,10 +62,11 @@ We extend our heartfelt thanks to all Pyronear volunteers and our SDIS partners
 
 Additionally, we express our gratitude to the DINUM for their financial and strategic support through the AIC, Etalab, and the Legal Service. Special thanks also go to the Mission Stratégie Prospective (MSP) for their guidance and collaboration.
 
-The Pyro-SDIS Subset contains **33,636 images**, including:
+Pyro-SDIS contains **12,378 images** (train 11,249, val 1,129), including:
 
-- **28,103 images with smoke**
-- **31,975 smoke instances**
+- **10,744 images with smoke**
+- **12,137 smoke instances**
+- **1,634 images without smoke**, false positives of the Pyronear system
 
 This dataset is formatted to be compatible with the Ultralytics YOLO framework, enabling efficient training of object detection models.
 
@@ -77,6 +78,7 @@ This dataset is formatted to be compatible with the Ultralytics YOLO framework,
 
 | Version | Year | Images | Notes |
 |---------|------|--------|-------|
+| v2.0 | 2026 | 12,378 | Images from 2023 to 2025, false positives added |
 | v1.0 | 2024 | 33,636 | Initial release |
 
 To load a specific version:
@@ -86,15 +88,21 @@ from datasets import load_dataset
 dataset = load_dataset("pyronear/pyro-sdis", revision="v1.0")
 ```
 
+Changes in v2.0:
+- At most 10 images per smoke sequence, so consecutive near duplicate frames are no longer all kept. v2.0 therefore has fewer images than v1.0.
+- 1,634 false positive images added, with empty annotations.
+- New partners and cameras, images from 2023 to 2025.
+- Smoke boxes use class `0` (v1.0 used `1`), matching `nc: 1` in `data.yaml`.
+
 
 ## Dataset Overview
 
 ### Contents
-The Pyro-SDIS Subset contains images and annotations for wildfire smoke detection. The dataset is structured with the following metadata for each image:
+Pyro-SDIS contains images and annotations for wildfire smoke detection. The dataset is structured with the following metadata for each image:
 
 - **Image Path**: File path to the image.
 - **Annotations**: YOLO-format bounding box annotations for smoke detection:
-  - `class_id`: Class label (e.g., smoke).
+  - `class_id`: always `0` (smoke). Images without smoke have empty annotations.
   - `x_center`, `y_center`: Normalized center coordinates of the bounding box.
   - `width`, `height`: Normalized width and height of the bounding box.
 - **Metadata**:
@@ -242,7 +250,8 @@ If you use this dataset, please cite:
 @dataset{pyro-sdis,
   author = {Pyronear Team},
   title = {Pyro-SDIS Dataset},
-  year = {2024},
+  year = {2026},
+  version = {2.0},
   publisher = {Hugging Face},
   url = {https://huggingface.co/datasets/pyronear/pyro-sdis}
 }