Created
March 19, 2019 10:32
-
-
Save vishal-keshav/a0a1c0b526a9fd3f0bf6356cac88a23d to your computer and use it in GitHub Desktop.
How to infer an image classification with pre-trained frozen tensorflow pb file
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import tensorflow as tf | |
from tensorflow.python.platform import gfile | |
import numpy as np | |
from imagenet_classes import class_names | |
from scipy.misc import imread, imresize | |
dir_name = 'mobilenet_v1_1.0_224' | |
with tf.Graph().as_default() as graph: | |
with tf.Session() as sess: | |
with gfile.FastGFile(dir_name + "/mobilenet_v1_1.0_224_frozen.pb") as f: | |
file = 'file1.jpg' | |
input = imread(file, mode='RGB') | |
input = imresize(input, (224, 224)).reshape(1, 224, 224, 3).astype(float) | |
input/=127.5 | |
input-=1. | |
graph_def = tf.GraphDef() | |
graph_def.ParseFromString(f.read()) | |
sess.graph.as_default() | |
tf.import_graph_def(graph_def, input_map=None, return_elements=None, | |
name="", op_dict=None, producer_op_list=None) | |
for op in graph.get_operations(): | |
print("Operation Name :" + op.name) | |
print("Tensor Stats :" + str(op.values())) | |
l_input = graph.get_tensor_by_name('input:0') | |
intermediate = graph.get_tensor_by_name('MobilenetV1/MobilenetV1/Conv2d_0/Relu6:0') | |
l_output = graph.get_tensor_by_name('MobilenetV1/Predictions/Reshape_1:0') | |
tf.global_variables_initializer() | |
inter_out = sess.run(intermediate, feed_dict = {l_input : input}) | |
print(inter_out) | |
op_prob = sess.run(l_output, feed_dict = {l_input : input}) | |
preds = (np.argsort(op_prob[0])[::-1])[0:5] | |
for p in preds: | |
print(class_names[p-1], op_prob[0][p]) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment