search

Hanson0910 / RetinaTrack

The pytorch implement of the retina track original paper link: https://arxiv.org/abs/2003.13870
Apache License 2.0
45 stars 9 forks source link

task_shared网络写得有一点小问题 #6

Open Kingpopen opened 2 years ago

Kingpopen commented 2 years ago

task_shared网络写得好像有些小问题，应该将peranchor_feature的定义放到第一个循环里面， 不然每一次peranchor_feature都是在上一个peranchor_feature的基础上累加。

def task_shared (inp, oup, stride = 1, leaky = 0, m1 = 3,anchorNum = 3) :

allanchor_feature =nn.ModuleList()
for i in range(anchorNum):
    peranchor_feature =  nn.ModuleList()
    for j in range(m1):
        peranchor_feature.append(conv_bn(inp,oup,stride,leaky))
    allanchor_feature.append(nn.Sequential(*peranchor_feature))
return allanchor_feature
Hanson0910 commented 2 years ago

task_shared网络写得好像有些小问题，应该将peranchor_feature的定义放到第一个循环里面， 不然每一次peranchor_feature都是在上一个peranchor_feature的基础上累加。

def task_shared (inp, oup, stride = 1, leaky = 0, m1 = 3,anchorNum = 3) :

allanchor_feature =nn.ModuleList()
for i in range(anchorNum):
    peranchor_feature =  nn.ModuleList()
    for j in range(m1):
        peranchor_feature.append(conv_bn(inp,oup,stride,leaky))
    allanchor_feature.append(nn.Sequential(*peranchor_feature))
return allanchor_feature

感谢提醒！