VGG训练报错:Errors may have originated from an input operation.

最新推荐文章于 2023-11-06 21:25:22 发布

福尔摩斯帅双

最新推荐文章于 2023-11-06 21:25:22 发布

阅读量1.4w

点赞数 2

文章标签：神经网络 tensorflow

本文链接：https://blog.youkuaiyun.com/qq_42026651/article/details/108683175

版权

在使用自定义数据集训练VGG时遇到错误：'Errors may have originated from an input operation.'。错误源于节点'conv0/Conv1/Conv2D'的输入操作。解决方法是将训练集图片的尺寸调整为224x224像素，以匹配VGG和ResNet网络的要求。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

利用自己做的数据集训练VGG时,出现报错信息:

Errors may have originated from an input operation.
Input Source operations connected to node conv0/Conv1/Conv2D:
IteratorGetNext (defined at train_resnet.py:95)

Input Source operations connected to node conv0/Conv1/Conv2D:
IteratorGetNext (defined at train_resnet.py:95)

Original stack trace for ‘conv0/Conv1/Conv2D’:
File “train_resnet.py”, line 103, in
logits = model.build(images, 2)
File “/home/nuaa301/yangshuang/MathModel/ResNet/model_resnet.py”, line 87, in build
net = self._conv_layer(input, 16, tf.nn.relu, name=‘Conv1’) # 48x48
File “/home/nuaa301/yangshuang/MathModel/ResNet/model_resnet.py”, line 14, in _conv_layer
name=name)
File “/home/nuaa301/anaconda3/envs/yang/lib/python3.6/site-packages/tensorflow/python/util/deprecation.py”, line 324, in new_func
return func(*args, **kwargs)
File “/home/nuaa301/anacon