Datasets:
fix the depth description
Browse files
README.md
CHANGED
|
@@ -65,33 +65,34 @@ python SpaceSense-Toolkit/convert/airsim_to_yolo.py --raw-data data_example --ou
|
|
| 65 |
|
| 66 |
| Modality | Format | Unit / Range | Description |
|
| 67 |
|----------|--------|-------------|-------------|
|
| 68 |
-
| RGB | PNG (1024x1024) | 8-bit color | Scene rendering |
|
| 69 |
-
| Depth |
|
| 70 |
-
| Semantic Segmentation | PNG (1024x1024
|
| 71 |
-
| LiDAR Point Cloud | ASC (x
|
| 72 |
-
| 6-DoF Pose | CSV | meters + Hamilton quaternion (w,x,y,z) |
|
| 73 |
|
| 74 |
## Coordinate System & Units
|
| 75 |
|
| 76 |
| Item | Convention |
|
| 77 |
|------|-----------|
|
| 78 |
-
| Camera Frame | X-forward, Y-right, Z-down (right-
|
| 79 |
| World Frame | AirSim NED, target spacecraft fixed at origin |
|
| 80 |
| Quaternion | Hamilton convention: w + xi + yj + zk |
|
| 81 |
| Euler Angles | ZYX intrinsic (Yaw-Pitch-Roll) |
|
| 82 |
| Position | meters (m), 6 decimal places |
|
| 83 |
-
| Depth Map | millimeters (mm),
|
| 84 |
-
| LiDAR | meters (m), .asc format (x
|
| 85 |
-
| Timestamp | YYYYMMDDHHMMSSmmm |
|
| 86 |
|
| 87 |
## Sensor Configuration
|
| 88 |
|
| 89 |
-
### Camera (cam0)
|
| 90 |
|
| 91 |
- Resolution: 1024 x 1024
|
| 92 |
- FOV: 50 degrees
|
| 93 |
-
- Image types captured: RGB (type 0), Segmentation (type 5), Depth (type 2)
|
| 94 |
- TargetGamma: 1.0
|
|
|
|
| 95 |
|
| 96 |
### LiDAR
|
| 97 |
|
|
@@ -99,7 +100,7 @@ python SpaceSense-Toolkit/convert/airsim_to_yolo.py --raw-data data_example --ou
|
|
| 99 |
- Channels: 256
|
| 100 |
- Vertical FOV: -20 to +20 degrees
|
| 101 |
- Horizontal FOV: -20 to +20 degrees
|
| 102 |
-
- Data frame: SensorLocalFrame
|
| 103 |
|
| 104 |
## Data Split (Zero-shot / OOD)
|
| 105 |
|
|
@@ -113,21 +114,34 @@ The training and validation sets contain **completely non-overlapping satellite
|
|
| 113 |
|
| 114 |
## Data Organization
|
| 115 |
|
| 116 |
-
Each `.tar.gz` file in the `raw/` folder contains data for one satellite
|
| 117 |
|
| 118 |
-
|
| 119 |
-
|
|
|
|
|
|
|
| 120 |
├── approach_front/
|
| 121 |
-
│ ├──
|
| 122 |
-
│ ├── depth/
|
| 123 |
-
│ ├──
|
| 124 |
-
│ ├── lidar/
|
| 125 |
-
│ └──
|
| 126 |
├── approach_back/
|
| 127 |
├── orbit_xy/
|
| 128 |
└── ...
|
| 129 |
```
|
| 130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
## Semantic Class Definitions
|
| 132 |
|
| 133 |
| Class ID | Name | Description |
|
|
@@ -154,4 +168,4 @@ This dataset is released under the [CC-BY-NC-4.0](https://creativecommons.org/li
|
|
| 154 |
year={2026},
|
| 155 |
url={https://arxiv.org/abs/2603.09320}
|
| 156 |
}
|
| 157 |
-
```
|
|
|
|
| 65 |
|
| 66 |
| Modality | Format | Unit / Range | Description |
|
| 67 |
|----------|--------|-------------|-------------|
|
| 68 |
+
| RGB | PNG (1024x1024) | 8-bit color | Scene rendering from `cam0` |
|
| 69 |
+
| Depth | NPZ (compressed) | int32, millimeters (`depth` key); background = 10,000 m | Per-pixel metric depth exported from AirSim `DepthPerspective` |
|
| 70 |
+
| Semantic Segmentation | PNG (1024x1024, RGB color mask) | color-coded part labels | Component-level segmentation mask; toolkit maps colors to 7 semantic classes |
|
| 71 |
+
| LiDAR Point Cloud | ASC (x,y,z per line) | meters, 3 decimal places | Sparse 3D point cloud |
|
| 72 |
+
| 6-DoF Pose | CSV | meters + Hamilton quaternion (w,x,y,z) | Target-spacecraft pose in camera frame, plus service/camera pose in world frame |
|
| 73 |
|
| 74 |
## Coordinate System & Units
|
| 75 |
|
| 76 |
| Item | Convention |
|
| 77 |
|------|-----------|
|
| 78 |
+
| Camera Frame | X-forward, Y-right, Z-down (FRD, right-handed) |
|
| 79 |
| World Frame | AirSim NED, target spacecraft fixed at origin |
|
| 80 |
| Quaternion | Hamilton convention: w + xi + yj + zk |
|
| 81 |
| Euler Angles | ZYX intrinsic (Yaw-Pitch-Roll) |
|
| 82 |
| Position | meters (m), 6 decimal places |
|
| 83 |
+
| Depth Map | millimeters (mm), stored as compressed `.npz`; deep-space background = 10,000 m |
|
| 84 |
+
| LiDAR | meters (m), `.asc` format (`x,y,z`), 3 decimal places |
|
| 85 |
+
| Timestamp | `YYYYMMDDHHMMSSmmm` |
|
| 86 |
|
| 87 |
## Sensor Configuration
|
| 88 |
|
| 89 |
+
### Camera (`cam0`)
|
| 90 |
|
| 91 |
- Resolution: 1024 x 1024
|
| 92 |
- FOV: 50 degrees
|
| 93 |
+
- Image types captured: RGB (`Scene`, type 0), Segmentation (type 5), Depth (`DepthPerspective`, type 2)
|
| 94 |
- TargetGamma: 1.0
|
| 95 |
+
- Mounting offset: 1 m in front of the service spacecraft body origin
|
| 96 |
|
| 97 |
### LiDAR
|
| 98 |
|
|
|
|
| 100 |
- Channels: 256
|
| 101 |
- Vertical FOV: -20 to +20 degrees
|
| 102 |
- Horizontal FOV: -20 to +20 degrees
|
| 103 |
+
- Data frame: `SensorLocalFrame`
|
| 104 |
|
| 105 |
## Data Split (Zero-shot / OOD)
|
| 106 |
|
|
|
|
| 114 |
|
| 115 |
## Data Organization
|
| 116 |
|
| 117 |
+
Each `.tar.gz` file in the `raw/` folder contains data for one satellite. The toolkit expects trajectory folders containing `image/`, `seg/`, `depth/`, `lidar/`, and `pose_ground_truth.csv`.
|
| 118 |
|
| 119 |
+
Typical layout:
|
| 120 |
+
|
| 121 |
+
```text
|
| 122 |
+
<satellite_name>/ or <timestamp>_<satellite_name>/
|
| 123 |
├── approach_front/
|
| 124 |
+
│ ├── image/ # RGB images (.png)
|
| 125 |
+
│ ├── depth/ # Depth maps (.npz, int32, mm, key = depth)
|
| 126 |
+
│ ├── seg/ # Semantic masks (.png, RGB color encoding)
|
| 127 |
+
│ ├── lidar/ # Point clouds (.asc)
|
| 128 |
+
│ └── pose_ground_truth.csv
|
| 129 |
├── approach_back/
|
| 130 |
├── orbit_xy/
|
| 131 |
└── ...
|
| 132 |
```
|
| 133 |
|
| 134 |
+
Some HuggingFace exports may contain one additional nesting level:
|
| 135 |
+
|
| 136 |
+
```text
|
| 137 |
+
<satellite_name>/
|
| 138 |
+
└── <satellite_name>/
|
| 139 |
+
├── approach_front/
|
| 140 |
+
├── approach_back/
|
| 141 |
+
├── orbit_xy/
|
| 142 |
+
└── ...
|
| 143 |
+
```
|
| 144 |
+
|
| 145 |
## Semantic Class Definitions
|
| 146 |
|
| 147 |
| Class ID | Name | Description |
|
|
|
|
| 168 |
year={2026},
|
| 169 |
url={https://arxiv.org/abs/2603.09320}
|
| 170 |
}
|
| 171 |
+
```
|