RepVGG模型加载

[OpenCDR]

您好，我尝试把RepVGG的backbone加到YOLOv5中，但现在报错信息如下：
Traceback (most recent call last):
File "/home/xxx/Desktop/flexible-yolov5/scripts/train.py", line 531, in
train(hyp, opt, device, tb_writer)
File "/home/xxx/Desktop/flexible-yolov5/scripts/train.py", line 91, in train
model = Model(opt.cfg).to(device) # create
File "/home/xxx/Desktop/flexible-yolov5/od/models/model.py", line 26, in init
backbone_out = self.backbone.out_shape
File "/home/xxx/Softwares/anaconda/envs/welding/lib/python3.7/site-packages/torch/nn/modules/module.py", line 948, in getattr
type(self).name, name))
AttributeError: 'RepVGG' object has no attribute 'out_shape'
您知道应该怎么修改模型吗？

[Bobo-y]

在RepVGG 加一个字典，像这样的，构建FPN时需要传入backbone的每一层输出的channel
self.out_shape = {'C3_size': self.out_channels[-3],
'C4_size': self.out_channels[-2],
'C5_size': self.out_channels[-1]}

@OpenCDR

[OpenCDR]

好的谢谢您。