https://github.com/ultralytics/yolov3/blob/master/models.py

Suppress bounding boxes according to their IoUs and confidence scores. Args: bbox (array): Bounding boxes to be transformed. The shape is

:param bboxes: 假设有N个bbox的score大于score_threshold，那么bboxes的shape为(N, 6)，存储格式为(xmin, ymin, xmax, ymax, score, class) 其中(xmin, ymin, xmax, ym

Performs Rotate-Non-Maximum Suppression (RNMS) on inference results； @param prediction: size=(batch_size, num, [xywh,score,num_classe

Performs Rotate-Non-Maximum Suppression (RNMS) on inference results； @param prediction: size=(batch_size, num, [xywh,score,num_classe

:param filename: ground truth file to parse :return: all instances in a picture

This function disables printing when not in master process

ap = voc_ap(rec, prec, [use_07_metric]) Compute VOC AP given precision and recall. If use_07_metric is true, uses the VOC 07 11 point met

https://arxiv.org/abs/1902.09630 boxes1(boxes2)' shape is [..., (x,y,w,h)].The size is for original image.

`Resnet50` is a function that returns a `Backbone` object with the following parameters: - `name`: "resnet50" - `pretrained`: False - `di

a cosine decay scheduler about steps, not epochs. :param optimizer: ex. optim.SGD :param T_max: max steps, and steps=epochs