tf-yolov4 1.0.1

YOLOv4 implementation with Tensorflow 2

tf-yolov4

YOLOv4 implementation with Tensorflow 2.

Example

import numpy as np
import PIL.Image
import yolov4

# Default: num_classes=80
yo = yolov4.YOLOv4(num_classes=80)

# Default: weights_path=None
# num_classes=80 and weights_path=None: Pre-trained COCO model will be loaded.
# num_classes!=80 and weights_path=None: Pre-trained backbone and SPP model will be loaded.
# Otherwise: User-defined weight file will be loaded.
yo.load_weights(weights_path=None)

img = np.array(PIL.Image.open('./data/sf.jpg'))

# The image with predicted bounding-boxes is created if `debug=True`
boxes, classes, scores = yo.predict(img, debug=True)

TODO

• Prediction
• Load Darknet weight file
• Pre-trained model
• Basic training function and Loss definition
• Label-smoothed BCE loss
• c-IoU loss
• Training data augmentation