DetHub: Object Detection Model Hub

Installation

pip install dethub

Yolov5 Object Prediction and Visualization

from dethub.model import Yolov5
from dethub.visualize import show

detection_model = Yolov5(model_path= "yolov5s.pt", device="cpu", confidence_threshold=0.5, image_size=640)
show("highway1.jpg", detection_model)

Torchvision Object Prediction and Visualization

from dethub.model import TorchVision
from dethub.visualize import show

detection_model = TorchVision(model_path= "dethub/models/torchvision/fasterrcnn_resnet50_fpn.pth", device="cpu", confidence_threshold=0.5, image_size=640)
show("highway1.jpg", detection_model)

TensorflowHub Object Prediction and Visualization

from dethub.model import TensorflowHub
from dethub.visualize import show

detection_model = TensorflowHub(model_path= "https://tfhub.dev/tensorflow/efficientdet/d3/1", device="cpu", confidence_threshold=0.5, image_size=640)
show("highway1.jpg", detection_model)

Contributing

Before opening a PR:

• Reformat with black and isort:

black . --config pyproject.toml
isort .

References:

• SAHI
• YOLOX
• Mcvarer