大佬大佬，请问要怎么样解决？
使用TPH训练，使用检测命令加载该训练权重时 python detect.py --weights runs_swin/exp/weights/best.pt 出现报错
Fusing layers...
Traceback (most recent call last):
File "detect.py", line 350, in
detect(kld_flag=opt.kld)
File "detect.py", line 54, in detect
model = attempt_load(weights, map_location=device) # load FP32 model
File "/content/drive//YOLOv5_OBB_KLD/models/experimental.py", line 150, in attempt_load
model.append(torch.load(w, map_location=map_location)['model'].float().fuse().eval()) # load FP32 model
File "/content/drive//YOLOv5_OBB_KLD/models/yolo_new.py", line 165, in fuse
if isinstance(m, (Conv, DWConv)) and hasattr(m, 'bn'): # 如果函数层名为Conv标准卷积层，且同时 层中包含‘bn’属性名
TypeError: isinstance() arg 2 must be a type or tuple of types
models/common_new.py 中
class Conv(nn.Module):中将 def forward_fuse(self, x) 函数名改成 与models/common.py中Conv一样的fuseforward的名字。
YOLOv5_OBB_KLD/models/yolo_new.py 中
改成
for m in self.model.modules():
if type(m) is Conv and hasattr(m, 'bn'): # 如果函数层名为Conv标准卷积层，且同时 层中包含‘bn’属性名
m._non_persistent_buffers_set = set() # pytorch 1.6.0 compatability
m.conv = fuse_conv_and_bn(m.conv, m.bn) # update conv
delattr(m, 'bn') # remove batchnorm 将'bn'属性删除
m.forward = m.fuseforward # update forward
self.info()
哇谢谢大佬回复，重新git pull了代码，又产生如下问题：
（1）train时未开启kld，还有角度信息angle
Epoch gpu_mem box obj cls angle total targets img_size
95/99 7.28G 0.8553 0.04603 0.01225 0.2448 1.158 101 1024: 100% 11/11 [00:10<00:00, 1.04it/s]
train时开启kld，每个epoch的angle为0
Epoch gpu_mem box obj cls angle total targets img_size
91/99 7.29G 0.4632 0.06264 0.01249 0 0.5384 62 1024: 100% 11/11 [00:08<00:00, 1.27it/s]
（2）detect时，使用上面tph.yaml产生的权重报错
Fusing layers...
Model Summary: 382 layers, 4.64237e+07 parameters, 0 gradients
Traceback (most recent call last):
File "detect.py", line 350, in
detect(kld_flag=opt.kld)
File "detect.py", line 96, in detect
_ = model(img.half() if half else img) if device.type != 'cpu' else None # run once
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(*input, **kwargs)
File "/content/drive/My Drive/YOLOv5_OBB_KLD/models/yolo_new.py", line 122, in forward
return self.forward_once(x, profile) # single-scale inference, train
File "/content/drive/My Drive/YOLOv5_OBB_KLD/models/yolo_new.py", line 140, in forward_once
x = m(x)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(*input, **kwargs)
File "/content/drive/My Drive/YOLOv5_OBB_KLD/models/common_new.py", line 480, in forward
return self.cv3(torch.cat((self.m(self.cv1(x)), self.cv2(x)), dim=1))
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(*input, **kwargs)
File "/content/drive/My Drive/YOLOv5_OBB_KLD/models/common_new.py", line 432, in forward
x = self.tr(x)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(input, **kwargs)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/container.py", line 117, in forward
input = module(input)
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 722, in _call_impl
result = self.forward(input, **kwargs)
File "/content/drive/My Drive/YOLOv5_OBB_KLD/models/common_new.py", line 398, in forward
attn_windows = self.attn(x_windows, mask=attn_mask) # [nWB, MhMw, C]
File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 722, in call_impl
result = self.forward(*input, **kwargs)
File "/content/drive/My Drive/YOLOv5_OBB_KLD/models/common_new.py", line 319, in forward
x = (attn @ v).transpose(1, 2).reshape(B, N, C)
RuntimeError: Expected object of scalar type Float but got scalar type Half for argument #2 'mat2' in call to _th_bmm_out
求大佬不吝指教~~
1）train时未开启kld，使用的是CSL法，1类为1°，角度信息用分类来获取，angle loss 就是那180个类loss。开启kld之后，角度加上bbox 5参数，一起回归，loss就放在了bbox loss里面了。因为代码基础是根据CSL上改的，所以你开启KLD之后，angle loss 为0。
2）tph.yaml 中模型有用到 Tansformer编码块，我用的代码不支持使用float16。当你要用tansformer的时候，可以在detect 把 half 弄成Flase.
大佬，请问可以和6.0版合并吗，看介绍说6.0版在各方面提升了不少。
大佬，请问可以和6.0版合并吗，看介绍说6.0版在各方面提升了不少。
你看看最新上传的yolov5_obb的代码，具体实验指标还在测试，你可以先试试跑跑看。
好的，请问在yolov5_obb的utils中是不是少了部分文件呀
好的，请问在yolov5_obb的utils中是不是少了部分文件呀
重新上传了文件。
YOLOv5_OBB_KLD copied to clipboard
lx-cly
