|
--- |
|
license: mit |
|
tags: |
|
- ultralyticsplus |
|
- yolov8 |
|
- ultralytics |
|
- yolo |
|
- vision |
|
- instance-segmentation |
|
- pytorch |
|
- awesome-yolov8-models |
|
- solar-panels |
|
- pv-panels |
|
library_name: ultralytics |
|
library_version: 8.0.57 |
|
inference: false |
|
pipeline_tag: image-segmentation |
|
--- |
|
|
|
YOLOv8s trained on solar panels dataset https://app.roboflow.com/rzeszow-university-of-technology/solar-panels-seg/2 |
|
|
|
**Inference API:** [On Roboflow](https://app.roboflow.com/rzeszow-university-of-technology/solar-panels-seg/deploy/2) |
|
|
|
## Training results |
|
|
|
 |
|
|
|
*Labels:* |
|
 |
|
|
|
*Predictions:* |
|
 |
|
|
|
## How to use |
|
|
|
1. Instal ultralytics package. Follow their guide here: [Quickstart](https://docs.ultralytics.com/quickstart/) |
|
2. Clone this repository. |
|
3. Run inference |
|
```sh |
|
yolo segment predict model=best.pt imgsz=640 save=True source=image.png |
|
``` |
|
|