解决Errors may have originated from an input operation. Input Source operations connected to Conv2D

tf1报错：

Errors may have originated from an input operation.
Input Source operations connected to node depth_estimation/depth_net/preprocessing/r1/skipLayer/skipLayer_conv/Conv2D:
 depth_estimation/depth_net/preprocessing/256to128/bn_relu/Relu (defined at ./.././network/hourglass.py:92)	
 depth_estimation/depth_net/preprocessing/r1/skipLayer/skipLayer_conv/W/read (defined at ./.././network/hourglass.py:83)

Input Source operations connected to node depth_estimation/depth_net/preprocessing/r1/skipLayer/skipLayer_conv/Conv2D:
 depth_estimation/depth_net/preprocessing/256to128/bn_relu/Relu (defined at ./.././network/hourglass.py:92)	
 depth_estimation/depth_net/preprocessing/r1/skipLayer/skipLayer_conv/W/read (defined at ./.././network/hourglass.py:83)

解决：

1. gpustat看下进程，可能是上个session没有关闭。kill掉卡上所有进程
2. 在import语句后放这一段：

config = tf.compat.v1.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.compat.v1.Session(config=config)