metadata
license: apache-2.0
tags:
- UAV
- Deadwood
- detectron2
Model description
This model is trained for detecting both standing and fallen deadwood from UAV RGB images. More thorough description is available on https://mayrajeo.github.io/maskrcnn-deadwood.
Training data
The model was trained on expert-annotated deadwood data, acquired on during leaf-on season 16.-17.7.2019 from Hiidenportti, Sotkamo, Eastern-Finland. The ground resolution for the data varied between 3.9 and 4.4cm. In addition, the model was tested with data collected from Evo, Hämeenlinna, Southern-Finland, acquired on 11.7.2018. The data from Evo was used only for testing the models.
Metrics
| Metric | Hiidenportti | Evo |
|---|---|---|
| Patch AP50 | 0.704 | 0.519 |
| Patch AP | 0.366 | 0.252 |
| Patch AP groundwood | 0.326 | 0.183 |
| Patch AP uprightwood | 0.406 | 0.321 |
| Scene AP50 | 0.683 | 0.511 |
| Scene AP | 0.341 | 0.236 |
| Scene AP groundwood | 0.246 | 0.160 |
| Scene AP uprightwood | 0.436 | 0.311 |
How to use
from detectron2.engine import DefaultPredictor
from detectron2.config import get_cfg
from detectron2.data import build_detection_test_loader
import cv2
cfg = get_cfg()
cfg.merge_from_file(<path_to_model_config>)
cfg.OUTPUT_DIR = '<path_to_output>'
cfg.MODEL.WEIGHTS = '<path_to_weights>'
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.5 # score threshold for detections
predictor = DefaultPredictor(cfg)
img = cv2.imread('<path_to_image_patch>')
outputs = predictor(image)