Author: Lijun Yu
Email: lijun@lj-y.com
A submodule of object detection models.
- Mask R-CNN from Detectron2
- YOLOv5 official
- EfficientDet in PyTorch
import torch
from detectors import get_detector
detector_class = get_detector('Mask R-CNN') # Or YOLOv5, EfficientDet
detector = detector_class(gpu_id=0)
# images: a list of pytorch tensors as H x W x C[BGR] in [0, 256)
images = [torch.zeros(1080, 1920, 3)]
detections = detector(images)
for detection in detections:
# Detection attributes: object_types, image_boxes, detection_scoresSee actev_base.
See License.