pytorch训练权重转化为tensorflow模型的教训

模型构建时候有时候在工程量比较大的时候，不可避免使用迭代算法，迭代算法本身会让错误的追踪更加困难，因此掌握基本的框架之间的差异非常重要。以下均是在模型转换过程中出现的错误。

shuffle operation(shuffle 操作)

这个操作原本是用来将各个通道之间的信息进行打乱后，此时面临重要的问题就是，如果将通道打乱，在pytorch里面与tensorflow中间，两种通道排序是不一样的，是采用不同的通道数据排列进行的。

import tensorflow as tfdef channel_shuffle(x, groups):_, h, w, c = x.shape# c 通道进行划分x = tf.reshape(x, [-1, h, w, groups, c // groups])# 通道为基本单位的情况下，多group均采样重组x = tf.transpose(x, [0, 1, 2, 4, 3])  # 调整通道维度顺序# 混洗采样重组后再reshape变成之前的通道x = tf.reshape(x, [-1, h, w, c])return x# 示例张量
x = tf.random.normal((2, 3, 3, 8))
print("Original tensor:\n", x.numpy())# 进行通道混洗
shuffled_x = channel_shuffle(x, groups=2)
print("Shuffled tensor:\n", shuffled_x.numpy())

Pytorch下的GSBottleneck采用的Sequential具有差异

class GSBottleneck(nn.Module):# GS Bottleneck https://github.com/AlanLi1997/slim-neck-by-gsconvdef __init__(self, c1, c2, k=3, s=1):super().__init__()c_ = c2 // 2# for lightingself.conv_lighting = nn.Sequential(GSConv(c1, c_, 1, 1),GSConv(c_, c2, 1, 1, act=False))# for receptive fieldself.conv = nn.Sequential(GSConv(c1, c_, 3, 1),GSConv(c_, c2, 3, 1, act=False))self.shortcut = nn.Identity()def forward(self, x):return self.conv_lighting(x)

我遇到的坑为：

class TFGSBottleneck(tf.keras.layers.Layer):# GS Bottleneck https://github.com/AlanLi1997/slim-neck-by-gsconvdef __init__(self, c1, c2, k=3, s=1,w=None):super().__init__()c_ = c2 // 2# example# self.cv1 = TFConv(c1, c_, 1, 1, w=w.cv1)# self.cv2 = TFConv(c_, c2, 3, 1, g=g, w=w.cv2)self.conv_lighting = tf.keras.Sequential(TFGSConv(c1, c_, 1, 1,w=w.conv_lighting[0]),TFGSConv(c_, c2, 1, 1, act=False, w=w.conv_lighting[1]))# for receptive fieldself.conv = tf.keras.Sequential(TFGSConv(c1, c_, 3, 1,w=w.conv[0]),TFGSConv(c_, c2, 3, 1, act=False,w=w.conv[1]))self.shortcut = tf.keras.layers.Lambda(lambda x: x)def call(self, x):print("TFGSBottleneck input: ",x.shape)print("TFGSBottleneck output: ", self.conv_lighting(x).shape)return self.conv_lighting(x)

有以下错误

Traceback (most recent call last):File "D:\TEST\yolov5\models\tf.py", line 1078, in <module>main(opt)File "D:\TEST\yolov5\models\tf.py", line 1073, in mainrun(**vars(opt))File "D:\TEST\yolov5\models\tf.py", line 1044, in run_ = tf_model.predict(im)  # inference^^^^^^^^^^^^^^^^^^^^File "D:\TEST\yolov5\models\tf.py", line 922, in predictx = m(x)  # run^^^^File "C:\Users\Zhuliang\.conda\envs\exportyolo2\Lib\site-packages\keras\utils\traceback_utils.py", line 70, in error_handlerraise e.with_traceback(filtered_tb) from NoneFile "D:\TEST\yolov5\models\tf.py", line 383, in callm_x1 = self.m(x1)^^^^^^^^^^File "D:\TEST\yolov5\models\tf.py", line 361, in callprint("TFGSBottleneck output: ", self.conv_lighting(x).shape)^^^^^^^^^^^^^^^^^^^^^
ValueError: Exception encountered when calling layer 'tfgs_bottleneck' (type TFGSBottleneck).name for name_scope must be a string.Call arguments received by layer 'tfgs_bottleneck' (type TFGSBottleneck):• x=tf.Tensor(shape=(1, 136, 136, 8), dtype=float32)

解决方案为

错误地将两个层作为位置参数传递给tf.keras.Sequential构造函数，导致第二个参数被误解为name参数，而name必须是一个字符串，但用户传递了一个层实例。这导致在调用层时，TensorFlow无法正确创建name_scope，从而引发错误。解决方法是把这两个层放在一个列表中，作为Sequential构造函数的第一个参数。

要解决此错误，需要将传递给tf.keras.Sequential的层放在列表中，确保它们被正确解析为层序列而不是其他参数。以下是修改后的代码：

tf.keras.Sequential与tensorflow中的pytorch

class TFGSBottleneck(tf.keras.layers.Layer):# GS Bottleneck https://github.com/AlanLi1997/slim-neck-by-gsconvdef __init__(self, c1, c2, k=3, s=1, w=None):super().__init__()c_ = c2 // 2# 使用列表包裹层以正确传递self.conv_lighting = tf.keras.Sequential([TFGSConv(c1, c_, 1, 1, w=w.conv_lighting[0]),TFGSConv(c_, c2, 1, 1, act=False, w=w.conv_lighting[1])])self.conv = tf.keras.Sequential([TFGSConv(c1, c_, 3, 1, w=w.conv[0]),TFGSConv(c_, c2, 3, 1, act=False, w=w.conv[1])])self.shortcut = tf.keras.layers.Lambda(lambda x: x)def call(self, x):return self.conv_lighting(x)