Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -37,4 +37,30 @@ task_categories:
|
|
| 37 |
- object-detection
|
| 38 |
size_categories:
|
| 39 |
- n<1K
|
| 40 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
- object-detection
|
| 38 |
size_categories:
|
| 39 |
- n<1K
|
| 40 |
+
---
|
| 41 |
+
|
| 42 |
+
The **IndustrialLateralLoads** dataset is divided into a single split: `loads`. This split contains the following:
|
| 43 |
+
|
| 44 |
+
- A folder with all the images: `imgs`
|
| 45 |
+
- A folder containing a `.txt` file for each image, where each row corresponds to a load with its bounding box annotation: `annotations`
|
| 46 |
+
|
| 47 |
+
---
|
| 48 |
+
|
| 49 |
+
**Remark:** Each row in a `.txt` file follows this format:
|
| 50 |
+
|
| 51 |
+
`<class_name> <instance_id> <x_min> <y_min> <weight> <height>`,
|
| 52 |
+
|
| 53 |
+
where the field `<instance_id>` field is not relevant in the currently published dataset, as it is intended for segmentation.
|
| 54 |
+
|
| 55 |
+
---
|
| 56 |
+
|
| 57 |
+
### Install Hugging Face datasets package:
|
| 58 |
+
```sh
|
| 59 |
+
pip install datasets
|
| 60 |
+
```
|
| 61 |
+
|
| 62 |
+
### Download the dataset:
|
| 63 |
+
```python
|
| 64 |
+
from datasets import load_dataset
|
| 65 |
+
dataset = load_dataset("jjldo21/IndustrialLateralLoads")
|
| 66 |
+
```
|