Commit ·
c3b124b
1
Parent(s): 6405517
add info.json, datacard and correct information in readme
Browse files- DATASET_CARD.md +40 -0
- README.md +1 -0
- info.json +14 -0
DATASET_CARD.md
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# PiPER robot teaching episodes
|
| 2 |
+
|
| 3 |
+
Short dataset card for the PiPER robot teaching episodes dataset.
|
| 4 |
+
|
| 5 |
+
Description
|
| 6 |
+
|
| 7 |
+
Versioning
|
| 8 |
+
|
| 9 |
+
License
|
| 10 |
+
|
| 11 |
+
How to use
|
| 12 |
+
|
| 13 |
+
Contact
|
| 14 |
+
|
| 15 |
+
# PiPER Robot Teaching Episodes
|
| 16 |
+
|
| 17 |
+
Short dataset card for the PiPER robot teaching episodes dataset.
|
| 18 |
+
|
| 19 |
+
Description
|
| 20 |
+
- PiPER robot teaching episodes dataset containing HDF5 recordings, per-episode JSON metadata, and image folders captured from two cameras:
|
| 21 |
+
- `observation.images.table_cam` (cropped to 800x720)
|
| 22 |
+
- `observation.images.wrist_cam` (vertically flipped)
|
| 23 |
+
|
| 24 |
+
Summary
|
| 25 |
+
- **Total Episodes**: 13
|
| 26 |
+
- **Unique Tasks**: 12
|
| 27 |
+
- **Total Size (approx.)**: ~6.2 GB
|
| 28 |
+
- **Recording Date**: November 2025
|
| 29 |
+
|
| 30 |
+
Versioning
|
| 31 |
+
- `info.json` at the repo root contains `codebase_version: v0.1` and the repository should be tagged with `v0.1` on the Hugging Face dataset. Creating the tag that matches `codebase_version` is the recommended way to ensure downstream tooling detects the dataset version.
|
| 32 |
+
|
| 33 |
+
License
|
| 34 |
+
- MIT
|
| 35 |
+
|
| 36 |
+
How to use
|
| 37 |
+
- Download the dataset from the Hugging Face Hub and parse the HDF5 files according to your data loader. Image folders contain PNG frames named like `frame_XXXXX.png` under `*_images/observation.images.*`.
|
| 38 |
+
|
| 39 |
+
Contact
|
| 40 |
+
- Dataset author: `charithmunasinghe`
|
README.md
CHANGED
|
@@ -24,6 +24,7 @@ This dataset contains teleoperation demonstrations recorded using the PiPER robo
|
|
| 24 |
- **Total Size**: ~6.2 GB
|
| 25 |
- **Cameras**: Dual camera setup (table_cam + wrist_cam)
|
| 26 |
- **Recording Date**: November 2024
|
|
|
|
| 27 |
|
| 28 |
## Tasks Included
|
| 29 |
|
|
|
|
| 24 |
- **Total Size**: ~6.2 GB
|
| 25 |
- **Cameras**: Dual camera setup (table_cam + wrist_cam)
|
| 26 |
- **Recording Date**: November 2024
|
| 27 |
+
- **Recording Date**: November 2025
|
| 28 |
|
| 29 |
## Tasks Included
|
| 30 |
|
info.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"name": "PiPER robot teaching episodes",
|
| 3 |
+
"description": "PiPER robot teaching episodes dataset containing HDF5 episode files, per-episode JSON metadata, and image folders for two cameras (table_cam and wrist_cam).",
|
| 4 |
+
"codebase_version": "v0.1",
|
| 5 |
+
"version": "v0.1",
|
| 6 |
+
"format": "lerobot.dataset.v3",
|
| 7 |
+
"lerobot_format_version": "3.0",
|
| 8 |
+
"license": "MIT",
|
| 9 |
+
"authors": ["charithmunasinghe"],
|
| 10 |
+
"tasks": ["robot_perception", "object_recognition", "imitation_learning"],
|
| 11 |
+
"modalities": ["image", "hdf5", "json"],
|
| 12 |
+
"estimated_num_files": 9200,
|
| 13 |
+
"notes": "Include this file at the dataset root and tag the HF dataset repository with the matching `codebase_version` (v0.1)."
|
| 14 |
+
}
|