Why is there an InvalidArgumentError: You must feed a value for placeholder tensor 'dense_1_input'

idata · ‎07-31-2018

Hello team,

Not sure this is a duplicate from other issues, but I did not find any useful answer to this topic.

I am trying to use Keras to generate a trained model and to ultimately use it in the Movidius for inference.

For this, I am saving the model as TM_Model/tf_model.meta

After this I run the mvNCCompile command to generate a graph input file that could be used by the Movidius.

# Ubuntu 16.04.5 LTS

mvNCCompile   ./TF_Model/tf_model.meta   -in   dense_1/kernel    -on activation_1/Softmax    -o ./graph

mvNCCompile v02.00, Copyright @ Intel Corporation 2017

InvalidArgumentError (see above for traceback): You must feed a value for placeholder tensor 'dense_1_input' with dtype float and shape [?,3072]
         [[Node: dense_1_input = Placeholder[dtype=DT_FLOAT, shape=[?,3072], _device="/job:localhost/replica:0/task:0/device:CPU:0"]()]]

I am using the cats and dogs example as it is documented in this url:

https://pyimagesearch.com/2016/09/26/a-simple-neural-network-with-python-and-keras/

The following python script represents this model and saves the data to TF_Model/tf_model.meta

Applying the output of this script to the mvNCCompile command, generates the InvalidArgumentError message.

#  Run this application in a virtual environment
#  [venv]  python   catsdogs.py    --dataset   kaggle_dogs_vs_cats
#  kaggle_dogs_vs_cats directory contains the 25000 images (cats and dogs)

# import the necessary packages
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from keras.models import model_from_json
from keras.models import Sequential
from keras.layers import Activation
from keras.optimizers import SGD
from keras.layers import Dense
from keras.utils import np_utils
from keras import backend as K
from imutils import paths
import numpy as np
import argparse
import cv2
import os
import tensorflow as tf

def image_to_feature_vector(image, size=(32, 32)):
    return cv2.resize(image, size).flatten()

# construct the arguments and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-d", "--dataset", required=True,help="path to input dataset")
args = vars(ap.parse_args())

# grab the list of images that we'll be describing
print("[INFO] describing images...")
imagePaths = list(paths.list_images(args["dataset"]))

# initialize the data matrix and labels list
data = []
labels = []

# loop over the input images
for (i, imagePath) in enumerate(imagePaths):
    image = cv2.imread(imagePath)
    label = imagePath.split(os.path.sep)[-1].split(".")[0]
    features = image_to_feature_vector(image)
    data.append(features)
    labels.append(label)
    if i > 0 and i % 1000 == 0:
        print("[INFO] processed {}/{}".format(i, len(imagePaths)))

# encode the labels, converting them from strings to integers
le = LabelEncoder()
labels = le.fit_transform(labels)
data = np.array(data) / 255.0
labels = np_utils.to_categorical(labels, 2)

print("[INFO] constructing training/testing split...")
(trainData, testData, trainLabels, testLabels) = train_test_split(data, labels, test_size=0.25, random_state=42)

# define the architecture of the network
model = Sequential()
model.add(Dense(768, input_dim=3072, init="uniform", activation="relu"))
model.add(Dense(384, activation="relu", kernel_initializer="uniform"))
model.add(Dense(2))
model.add(Activation("softmax"))

# train the model using SGD
print("[INFO] compiling model...")
sgd = SGD(lr=0.01)
model.compile(loss="binary_crossentropy", optimizer=sgd, metrics=["accuracy"])

# epochs is set to 5 iso 50, to speed up a little.
model.fit(trainData, trainLabels, epochs=5, batch_size=128, verbose=1)

(loss, accuracy) = model.evaluate(testData, testLabels, batch_size=128, verbose=1)
print("[INFO] loss={:.4f}, accuracy: {:.4f}%".format(loss,accuracy * 100))

K.set_learning_phase(0)

# serialize model to JSON
print("[INFO] saving model in json format (model.json)")
model_json = model.to_json()
with open("model.json", "w") as json_file:
        json_file.write(model_json)

# serialize weights to HDF5
print("[INFO] saving weigths in h5 format (model.h5)")
model.save_weights("model.h5")

print("[INFO] Saving to TF_Model (TF_Model/tf_model)")
saver = tf.train.Saver()
sess = K.get_session()
saver.save(sess, "./TF_Model/tf_model")

Any ideas to get this solved are very welcome, thanks !

idata · ‎08-09-2018

I just got exactly the same problem. Hope someone can help.

idata · ‎09-15-2018

Did anybody figured out why this error is coming. I am also stuck in this

idata · ‎11-14-2018

Hi

Please let me know if there is solution for this problem. I am stuck at this problem.

GWang48 · ‎08-13-2019

For NCSDK 1.X, this solution works for me:

edit ncsdk source in

/usr/local/bin/ncsdk/Controllers/TensorFlowParser.py

line 1059, add a feed_dict to eval:

# desired_shape = node.inputs[1].eval()

desired_shape = node.inputs[1].eval(feed_dict={inputnode + ':0' : input_data})

Enjoy ;-)