Upload README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,72 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Pose Estimation Dataset
|
| 2 |
+
|
| 3 |
+
## Overview
|
| 4 |
+
|
| 5 |
+
This dataset is designed for pose estimation tasks, focusing on determining the position and orientation of an object in 3D space. The dataset includes images, masks, and labels for both training and validation, making it suitable for machine learning applications in 3D object tracking and computer vision. This dataset was generated using Duality.ai simulation software: FalconEditor. [Try FalconEditor today](https://falcon.duality.ai/secure/documentation?learnWelcome=true&sidebarMode=learn) to create data to be used for pose estimation on different objects.
|
| 6 |
+
|
| 7 |
+
## Dataset Structure
|
| 8 |
+
|
| 9 |
+
The dataset has the following structure:
|
| 10 |
+
```plaintext
|
| 11 |
+
pose_estimation_dataset/
|
| 12 |
+
|-- train/
|
| 13 |
+
| |-- images/
|
| 14 |
+
| | |-- 000000000.png
|
| 15 |
+
| | |-- 000000001.png
|
| 16 |
+
| | |-- ...
|
| 17 |
+
| |-- labels/
|
| 18 |
+
| | |-- 000000000.txt
|
| 19 |
+
| | |-- 000000001.txt
|
| 20 |
+
| | |-- ...
|
| 21 |
+
| |-- masks/
|
| 22 |
+
| |-- 000000000.png
|
| 23 |
+
| |-- 000000001.png
|
| 24 |
+
| |-- ...
|
| 25 |
+
|-- val/
|
| 26 |
+
|-- images/
|
| 27 |
+
| |-- 000000000.png
|
| 28 |
+
| |-- 000000001.png
|
| 29 |
+
| |-- ...
|
| 30 |
+
|-- labels/
|
| 31 |
+
| |-- 000000000.txt
|
| 32 |
+
| |-- 000000001.txt
|
| 33 |
+
| |-- ...
|
| 34 |
+
|-- masks/
|
| 35 |
+
|-- 000000000.png
|
| 36 |
+
|-- 000000001.png
|
| 37 |
+
|-- ...
|
| 38 |
+
```
|
| 39 |
+
|
| 40 |
+
### Components
|
| 41 |
+
1. **Images**: RGB images of the object in `.png` format.
|
| 42 |
+
2. **Labels**: Text files (`.txt`) containing 3D pose annotations. Each label file corresponds to an image and contains the following information:
|
| 43 |
+
- Three position values `[x, y, z]` representing the object's location in 3D space.
|
| 44 |
+
- Four quaternion values `[qx, qy, qz, qw]` representing the object's orientation in 3D space.
|
| 45 |
+
3. **Masks**: Binary masks (`.png`) highlighting the object’s silhouette in the image.
|
| 46 |
+
|
| 47 |
+
## Usage
|
| 48 |
+
|
| 49 |
+
To use this dataset, load the images, labels, and masks for your pose estimation pipeline. Ensure that the corresponding image, label, and mask files share the same base filename.
|
| 50 |
+
|
| 51 |
+
### Example
|
| 52 |
+
If you have `train/images/image_1.png`, the corresponding files will be:
|
| 53 |
+
- `train/labels/image_1.txt`
|
| 54 |
+
- `train/masks/image_1.png`
|
| 55 |
+
|
| 56 |
+
### Label Format
|
| 57 |
+
Each `.txt` label file contains a single line in the following format:
|
| 58 |
+
```plaintext
|
| 59 |
+
x y z qx qy qz qw
|
| 60 |
+
```
|
| 61 |
+
Example:
|
| 62 |
+
```plaintext
|
| 63 |
+
0.12 0.45 0.78 0.0 0.707 0.0 0.707
|
| 64 |
+
```
|
| 65 |
+
|
| 66 |
+
## Licensing
|
| 67 |
+
license: apache-2.0
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
## Acknowledgements
|
| 71 |
+
Include any acknowledgements or credits if applicable.
|
| 72 |
+
|