Datasets:
Translate dataset README to English and link pipeline
Browse files
README.md
CHANGED
|
@@ -4,47 +4,54 @@ license: mit
|
|
| 4 |
|
| 5 |
# SpatialSemanticDataset
|
| 6 |
|
| 7 |
-
SpatialSemanticDataset
|
| 8 |
|
| 9 |
-
|
| 10 |
|
| 11 |
-
|
| 12 |
|
| 13 |
-
|
|
|
|
|
|
|
| 14 |
| --- | --- |
|
| 15 |
-
| `metadata.csv` |
|
| 16 |
-
| `poses/<channel>/<video_id>/<scene_id>.csv` |
|
| 17 |
-
| `semantics/<channel>/<video_id>/<scene_id>.txt` |
|
| 18 |
|
| 19 |
-
|
| 20 |
|
| 21 |
-
##
|
| 22 |
|
| 23 |
-
###
|
| 24 |
|
| 25 |
-
|
| 26 |
-
-
|
| 27 |
-
|
|
|
|
|
|
|
|
|
|
| 28 |
|
| 29 |
-
###
|
| 30 |
|
| 31 |
-
|
|
| 32 |
| --- | --- |
|
| 33 |
-
| `source_t` |
|
| 34 |
-
| `source_frame` |
|
| 35 |
-
| `status` |
|
| 36 |
-
| `x`, `y`, `z` |
|
| 37 |
-
| `qx`, `qy`, `qz`, `qw` |
|
|
|
|
|
|
|
| 38 |
|
| 39 |
-
|
| 40 |
|
| 41 |
-
|
| 42 |
|
| 43 |
-
|
| 44 |
|
| 45 |
-
##
|
| 46 |
|
| 47 |
-
|
| 48 |
|
| 49 |
```python
|
| 50 |
import csv
|
|
@@ -74,8 +81,8 @@ print(f"Valid poses: {len(valid_poses)}")
|
|
| 74 |
print(annotations.split("\n\n", 1)[0])
|
| 75 |
```
|
| 76 |
|
| 77 |
-
|
| 78 |
|
| 79 |
-
##
|
| 80 |
|
| 81 |
-
|
|
|
|
| 4 |
|
| 5 |
# SpatialSemanticDataset
|
| 6 |
|
| 7 |
+
SpatialSemanticDataset is a dataset of first-person walking videos annotated with camera poses and English semantic descriptions. It organizes annotations by scene and supports research on visual navigation, camera trajectory and language alignment, and descriptions of observations and intended actions.
|
| 8 |
|
| 9 |
+
The dataset covers **12,079 scenes from 3,188 videos across 9 source channels**. This repository contains the scene index and annotations. The source videos are hosted separately and can be located by their YouTube video IDs.
|
| 10 |
|
| 11 |
+
The data processing pipeline is maintained in a separate repository. [SpatialSemanticPipeline](https://github.com/monsterdie12138/SpatialSemanticPipeline)
|
| 12 |
|
| 13 |
+
## Repository layout
|
| 14 |
+
|
| 15 |
+
| Path | Contents |
|
| 16 |
| --- | --- |
|
| 17 |
+
| `metadata.csv` | Scene index, with one row per scene |
|
| 18 |
+
| `poses/<channel>/<video_id>/<scene_id>.csv` | Timestamped camera positions, orientations, and pose status |
|
| 19 |
+
| `semantics/<channel>/<video_id>/<scene_id>.txt` | Timestamped English observations, inferences, and action descriptions |
|
| 20 |
|
| 21 |
+
Use `(channel, video_id, scene_id)` to join the index with the two annotation files. Scene IDs such as `scene_0000` are local to each video.
|
| 22 |
|
| 23 |
+
## Data fields
|
| 24 |
|
| 25 |
+
### Scene index
|
| 26 |
|
| 27 |
+
| Field | Meaning |
|
| 28 |
+
| --- | --- |
|
| 29 |
+
| `channel` | Source channel |
|
| 30 |
+
| `video_id` | YouTube video ID |
|
| 31 |
+
| `scene_id` | Scene identifier within the source video |
|
| 32 |
+
| `start_second`, `end_second` | Scene start and end times in the source video, in seconds |
|
| 33 |
|
| 34 |
+
### Camera poses
|
| 35 |
|
| 36 |
+
| Field | Meaning |
|
| 37 |
| --- | --- |
|
| 38 |
+
| `source_t` | Timestamp in the source video, in seconds |
|
| 39 |
+
| `source_frame` | Frame index in the source video |
|
| 40 |
+
| `status` | Pose status. Sample files contain `valid` rows with pose values and `bad` rows whose pose fields can be `NaN`. |
|
| 41 |
+
| `x`, `y`, `z` | Camera position components |
|
| 42 |
+
| `qx`, `qy`, `qz`, `qw` | Camera orientation quaternion, with the scalar component `qw` last |
|
| 43 |
+
|
| 44 |
+
The dataset files leave the position units, axis conventions, and coordinate relationships between scenes unspecified. Confirm these conventions before calculating physical distances or combining trajectories across scenes.
|
| 45 |
|
| 46 |
+
### Semantic annotations
|
| 47 |
|
| 48 |
+
Annotations are grouped by source-video timestamp in `[HH:MM:SS]` format. Each block contains an `Observation` describing what is visible, a `Latent Thought` expressing an inferred situation or intent in the first person, and an `Action` describing an action or plan.
|
| 49 |
|
| 50 |
+
Pose samples and semantic annotations can occur at different times. Join them by scene, then align them on the source-video timeline.
|
| 51 |
|
| 52 |
+
## Load a scene
|
| 53 |
|
| 54 |
+
This repository is private. Install `huggingface_hub` and use `hf auth login` or set the `HF_TOKEN` environment variable with a token that has read access. This example downloads the index and the two annotation files for its first scene.
|
| 55 |
|
| 56 |
```python
|
| 57 |
import csv
|
|
|
|
| 81 |
print(annotations.split("\n\n", 1)[0])
|
| 82 |
```
|
| 83 |
|
| 84 |
+
The dataset ships without predefined splits. When creating training and evaluation splits, group scenes by `(channel, video_id)` so that scenes from the same source video stay in the same split.
|
| 85 |
|
| 86 |
+
## License
|
| 87 |
|
| 88 |
+
The dataset card declares the MIT license. Source videos are subject to their own terms.
|