It can be tested with the model file generated with below python script: import tensorflow as tf import numpy as np import imageio in_img = imageio.imread('input.jpg') in_img = in_img.astype(np.float32)/255.0 in_data = in_img[np.newaxis, :] x = tf.placeholder(tf.float32, shape=[1, None, None, 3], name='dnn_in') z1 = 0.039 + x z2 = x + 0.042 z3 = z1 + z2 z4 = z3 - 0.381 z5 = z4 - x y = tf.math.maximum(z5, 0.0, name='dnn_out') sess=tf.Session() sess.run(tf.global_variables_initializer()) graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out']) tf.train.write_graph(graph_def, '.', 'image_process.pb', as_text=False) print("image_process.pb generated, please use \ path_to_ffmpeg/tools/python/convert.py to generate image_process.model\n") output = sess.run(y, feed_dict={x: in_data}) imageio.imsave("out.jpg", np.squeeze(output))
Showing the changes to tools/python/convert_from_tensorflow.py (1 of 4 files). Show all
Unified · Split
| @@ -71,7 +71,7 @@ class TFConverter: · 70 unmodified lines | ||
| 71 | 71 | self.conv2d_scope_names = set() |
| 72 | 72 | self.conv2d_scopename_inputname_dict = {} |
| 73 | 73 | self.op2code = {'Conv2D':1, 'DepthToSpace':2, 'MirrorPad':3, 'Maximum':4, 'MathBinary':5} |
| 74 | − self.mathbin2code = {'Sub':0} | |
| 74 | + self.mathbin2code = {'Sub':0, 'Add':1} | |
| 75 | 75 | self.mirrorpad_mode = {'CONSTANT':0, 'REFLECT':1, 'SYMMETRIC':2} |
| 76 | 76 | self.name_operand_dict = {} |
| 77 | 77 | |
| @@ -255,8 +255,7 @@ class TFConverter: · 177 unmodified lines | ||
| 255 | 255 | np.array([input_operand_index, output_operand_index], dtype=np.uint32).tofile(f) |
| 256 | 256 | |
| 257 | 257 | |
| 258 | − def dump_sub_to_file(self, node, f): | |
| 259 | − assert(node.op == 'Sub') | |
| 258 | + def dump_mathbinary_to_file(self, node, f): | |
| 260 | 259 | self.layer_number = self.layer_number + 1 |
| 261 | 260 | self.converted_nodes.add(node.name) |
| 262 | 261 | i0_node = self.name_node_dict[node.input[0]] |
| @@ -264,15 +263,13 @@ class TFConverter: · 1 unmodified line | ||
| 264 | 263 | np.array([self.op2code['MathBinary'], self.mathbin2code[node.op]], dtype=np.uint32).tofile(f) |
| 265 | 264 | if i0_node.op == 'Const': |
| 266 | 265 | scalar = i0_node.attr['value'].tensor.float_val[0] |
| 267 | − assert(i0_node.name.find('sub/x')) | |
| 268 | − np.array([1], dtype=np.uint32).tofile(f) | |
| 266 | + np.array([1], dtype=np.uint32).tofile(f) # broadcast: 1 | |
| 269 | 267 | np.array([scalar], dtype=np.float32).tofile(f) |
| 270 | − np.array([0], dtype=np.uint32).tofile(f) | |
| 268 | + np.array([0], dtype=np.uint32).tofile(f) # broadcast: 0 | |
| 271 | 269 | input_operand_index = self.add_operand(i1_node.name, Operand.IOTYPE_INPUT) |
| 272 | 270 | np.array([input_operand_index], dtype=np.uint32).tofile(f) |
| 273 | 271 | elif i1_node.op == 'Const': |
| 274 | 272 | scalar = i1_node.attr['value'].tensor.float_val[0] |
| 275 | − assert(i1_node.name.find('sub/y')) | |
| 276 | 273 | np.array([0], dtype=np.uint32).tofile(f) |
| 277 | 274 | input_operand_index = self.add_operand(i0_node.name, Operand.IOTYPE_INPUT) |
| 278 | 275 | np.array([input_operand_index], dtype=np.uint32).tofile(f) |
| @@ -309,7 +306,9 @@ class TFConverter: · 30 unmodified lines | ||
| 309 | 306 | elif node.op == 'Maximum': |
| 310 | 307 | self.dump_maximum_to_file(node, f) |
| 311 | 308 | elif node.op == 'Sub': |
| 312 | − self.dump_sub_to_file(node, f) | |
| 309 | + self.dump_mathbinary_to_file(node, f) | |
| 310 | + elif node.op == 'Add': | |
| 311 | + self.dump_mathbinary_to_file(node, f) | |
| 313 | 312 | |
| 314 | 313 | |
| 315 | 314 | def dump_operands_to_file(self, f): |