I am a newb in tensorflow. I succeeded in restoring the graph based on the ckpt file given from the link above. However, I am struggling with finding the name of the input and output nodes.
In my past project with tensorflow, I used .get_tensor_by_name(' '). I understand there might be many other ways to run the pre-trained model, but it would be much easier for me if I can simply use get_tensor_by_name().
Please HELP ME :((
dir = './log_pretrained/log_v1/'
sess = tf.Session(config=tf.ConfigProto(allow_soft_placement=True, log_device_placement=True))
saver = tf.train.import_meta_graph(dir + 'model.ckpt.meta')
saver.restore(sess, tf.train.latest_checkpoint(dir))
graph = tf.get_default_graph()
y = graph.get_tensor_by_name('fc3/BiasAdd:0')
my laptop doesnt have a gpu on it, so I set the allow_soft_placement=True
This is a part of my code that is relevant to this issue.. I have no clue for the input node, and I am guessing 'fc3/BiasAdd:0' to be the output node.. correct me if I am wrong.
Also, I believe the output node should have a shape of 40, which is the number of the classes being output. I do not know how the shape of the input node would be. The input is a pointcloud file (which seems to have over 2000 points for each).
Firstly, thank you for sharing the codes.
I am conveniently using a pre-trained model given in (https://github.com/charlesq34/frustum-pointnets/issues/17), to input some point clouds and see if the output classification is correct.
I am a newb in tensorflow. I succeeded in restoring the graph based on the ckpt file given from the link above. However, I am struggling with finding the name of the input and output nodes.
In my past project with tensorflow, I used .get_tensor_by_name(' '). I understand there might be many other ways to run the pre-trained model, but it would be much easier for me if I can simply use get_tensor_by_name().
Please HELP ME :((
This is a part of my code that is relevant to this issue.. I have no clue for the input node, and I am guessing 'fc3/BiasAdd:0' to be the output node.. correct me if I am wrong.
Also, I believe the output node should have a shape of 40, which is the number of the classes being output. I do not know how the shape of the input node would be. The input is a pointcloud file (which seems to have over 2000 points for each).
Please help me.
Thank you very much in advance..