关于AnchorGenerator的问题
Created by: shuxsu
class AnchorGenerator(object): op = fluid.layers.anchor_generator append_doc = True
def __init__(self,
stride=[16.0, 16.0],
anchor_sizes=[32, 64, 128, 256, 512],
aspect_ratios=[0.5, 1., 2.],
variance=[1., 1., 1., 1.]):
super(AnchorGenerator, self).__init__()
self.anchor_sizes = anchor_sizes
self.aspect_ratios = aspect_ratios
self.variance = variance
self.stride = stride
根据官方API文档所示 为RCNN算法生成anchor,输入的每一位产生N个anchor,N=size(anchor_sizes)size(aspect_ratios)。生成anchor的顺序首先是aspect_ratios循环,然后是anchor_sizes循环。 意思是当anchor_sizes为32,aspect_ratios=0.5时候,N=3232*0.5=512个吗? 这些锚点的xmin,xmax,ymin,ymax是根据什么生成的呢?而且这512个是按照长宽比[0.5, 1., 2.] 这种分布的吗