RetinaTrack
RetinaTrack copied to clipboard
Hanson0910
The pytorch implement of the retina track original paper link: https://arxiv.org/abs/2003.13870
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...
class SSH(nn.Module): def __init__(self, in_channel, out_channel): super(SSH, self).__init__() assert out_channel % 4 == 0 leaky = 0 if (out_channel
我在视频上测试效果很差，几乎目标都检测不到，这是为什么 
Metadata
Owner
Metadata
The pytorch implement of the retina track original paper link: https://arxiv.org/abs/2003.13870