WARNING:tensorflow:Input graph does not use tf.data.Dataset or contain a QueueRunner.
WinstonDeng opened this issue · comments
I'm trying to use your model with 2d coordinates as input, so that I build this function to replace your eval_input_fn(filename, barch_size). The return of my function is same with yours, i.e.
({'points3d': <tf.Tensor 'Reshape_1:0' shape=(32, 17, 3) dtype=float32>, 'points2d': <tf.Tensor 'Reshape:0' shape=(32, 17, 2) dtype=float32>}, <tf.Tensor 'Reshape_2:0' shape=(32,) dtype=int32>)
but never stop with warning : WARNING:tensorflow:Input graph does not use tf.data.Dataset or contain a QueueRunner. That means predict yields forever. This is probably a mistake.
Do you have any idea to solve it?
my code is here
‘’‘
@param : keypoints2DList is a list.
Shpae = 32173.
keypoints2DList[ i , : , :2 ] is 2d coordinates of frame i
This function aims to build model input dataset from 2d coordinates, which is to replace your
eval_input_fn
’‘’
def my_eval_input_fn(keypoints2DList):
features={
'points2d': tf.FixedLenFeature((), tf.float32, default_value=""),
'points3d': tf.FixedLenFeature((), tf.float32, default_value=""),
}
keypoints2DList=np.array(keypoints2DList)
features.update(points2d=tf.reshape(keypoints2DList[..., :2], [32, 17, 2]))
features.update(points3d=tf.reshape(keypoints2DList, [32, 17, 3]))
temp=[]
for _ in range(32):
temp.append(0)
t = features, tf.reshape(temp, [32, ])
return tuple(t)
Thanks for you interest but I guess this warning is out of the scope of this project. You might want to google it and see why your code is shooting a warning. I would point you to read tensorflow/tensorflow#11621.
Thanks again your interest and I am closing this issue.