README.md 23.2 KB
Newer Older
1
# TensorFlow-Slim image classification model library
2
3
4
5
6
7
8
9
10
11
12
13
14
15

[TF-slim](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/slim)
is a new lightweight high-level API of TensorFlow (`tensorflow.contrib.slim`)
for defining, training and evaluating complex
models. This directory contains
code for training and evaluating several widely used Convolutional Neural
Network (CNN) image classification models using TF-slim.
It contains scripts that will allow
you to train models from scratch or fine-tune them from pre-trained network
weights. It also contains code for downloading standard image datasets,
converting them
to TensorFlow's native TFRecord format and reading them in using TF-Slim's
data reading and queueing utilities. You can easily train any model on any of
these datasets, as we demonstrate below. We've also included a
16
[jupyter notebook](https://github.com/tensorflow/models/blob/master/slim/slim_walkthrough.ipynb),
17
which provides working examples of how to use TF-Slim for image classification.
18
For developing or modifying your own models, see also the [main TF-Slim page](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/slim).
19

Alex Kurakin's avatar
Alex Kurakin committed
20
21
22
23
24
25
26
## Contacts

Maintainers of TF-slim:

* Nathan Silberman,
  github: [nathansilberman](https://github.com/nathansilberman)
* Sergio Guadarrama, github: [sguada](https://github.com/sguada)
27
28
29
30
31

## Table of contents

<a href="#Install">Installation and setup</a><br>
<a href='#Data'>Preparing the datasets</a><br>
32
<a href='#Pretrained'>Using pre-trained models</a><br>
33
34
35
<a href='#Training'>Training from scratch</a><br>
<a href='#Tuning'>Fine tuning to a new task</a><br>
<a href='#Eval'>Evaluating performance</a><br>
36
37
<a href='#Export'>Exporting Inference Graph</a><br>
<a href='#Troubleshooting'>Troubleshooting</a><br>
38
39
40
41
42
43
44
45
46

# Installation
<a id='Install'></a>

In this section, we describe the steps required to install the appropriate
prerequisite packages.

## Installing latest version of TF-slim

Neal Wu's avatar
Neal Wu committed
47
48
49
TF-Slim is available as `tf.contrib.slim` via TensorFlow 1.0. To test that your
installation is working, execute the following command; it should run without
raising any errors.
50

51
52
53
```
python -c "import tensorflow.contrib.slim as slim; eval = slim.evaluation.evaluate_once"
```
54

55
## Installing the TF-slim image models library
56

57
58
59
60
61
To use TF-Slim for image classification, you also have to install
the [TF-Slim image models library](https://github.com/tensorflow/models/tree/master/slim),
which is not part of the core TF library.
To do this, check out the
[tensorflow/models](https://github.com/tensorflow/models/) repository as follows:
62

63
64
65
66
```bash
cd $HOME/workspace
git clone https://github.com/tensorflow/models/
```
67

68
69
70
71
This will put the TF-Slim image models library in `$HOME/workspace/models/slim`.
(It will also create a directory called
[models/inception](https://github.com/tensorflow/models/tree/master/inception),
which contains an older version of slim; you can safely ignore this.)
72

73
74
To verify that this has worked, execute the following commands; it should run
without raising any errors.
75
76

```
896066322's avatar
896066322 committed
77
cd $HOME/workspace/models/slim
78
python -c "from nets import cifarnet; mynet = cifarnet.cifarnet"
79
80
81
```


82
# Preparing the datasets
83
<a id='Data'></a>
84

85
86
As part of this library, we've included scripts to download several popular
image datasets (listed below) and convert them to slim format.
87

88
89
90
91
92
93
Dataset | Training Set Size | Testing Set Size | Number of Classes | Comments
:------:|:---------------:|:---------------------:|:-----------:|:-----------:
Flowers|2500 | 2500 | 5 | Various sizes (source: Flickr)
[Cifar10](https://www.cs.toronto.edu/~kriz/cifar.html) | 60k| 10k | 10 |32x32 color
[MNIST](http://yann.lecun.com/exdb/mnist/)| 60k | 10k | 10 | 28x28 gray
[ImageNet](http://www.image-net.org/challenges/LSVRC/2012/)|1.2M| 50k | 1000 | Various sizes
94

95
## Downloading and converting to TFRecord format
96

97
98
99
100
101
102
For each dataset, we'll need to download the raw data and convert it to
TensorFlow's native
[TFRecord](https://www.tensorflow.org/versions/r0.10/api_docs/python/python_io.html#tfrecords-format-details)
format. Each TFRecord contains a
[TF-Example](https://github.com/tensorflow/tensorflow/blob/r0.10/tensorflow/core/example/example.proto)
protocol buffer. Below we demonstrate how to do this for the Flowers dataset.
103
104

```shell
105
106
107
108
$ DATA_DIR=/tmp/data/flowers
$ python download_and_convert_data.py \
    --dataset_name=flowers \
    --dataset_dir="${DATA_DIR}"
109
110
111
112
113
114
115
```

When the script finishes you will find several TFRecord files created:

```shell
$ ls ${DATA_DIR}
flowers_train-00000-of-00005.tfrecord
116
...
117
118
flowers_train-00004-of-00005.tfrecord
flowers_validation-00000-of-00005.tfrecord
119
...
120
121
122
123
124
125
126
127
flowers_validation-00004-of-00005.tfrecord
labels.txt
```

These represent the training and validation data, sharded over 5 files each.
You will also find the `$DATA_DIR/labels.txt` file which contains the mapping
from integer labels to class names.

128
129
130
131
You can use the same script to create the mnist and cifar10 datasets.
However, for ImageNet, you have to follow the instructions
[here](https://github.com/tensorflow/models/blob/master/inception/README.md#getting-started).
Note that you first have to sign up for an account at image-net.org.
Neal Wu's avatar
Neal Wu committed
132
Also, the download can take several hours, and could use up to 500GB.
133
134


135
## Creating a TF-Slim Dataset Descriptor.
136

137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
Once the TFRecord files have been created, you can easily define a Slim
[Dataset](https://github.com/tensorflow/tensorflow/blob/r0.10/tensorflow/contrib/slim/python/slim/data/dataset.py),
which stores pointers to the data file, as well as various other pieces of
metadata, such as the class labels, the train/test split, and how to parse the
TFExample protos. We have included the TF-Slim Dataset descriptors
for
[Cifar10](https://github.com/tensorflow/models/blob/master/slim/datasets/cifar10.py),
[ImageNet](https://github.com/tensorflow/models/blob/master/slim/datasets/imagenet.py),
[Flowers](https://github.com/tensorflow/models/blob/master/slim/datasets/flowers.py),
and
[MNIST](https://github.com/tensorflow/models/blob/master/slim/datasets/mnist.py).
An example of how to load data using a TF-Slim dataset descriptor using a
TF-Slim
[DatasetDataProvider](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/slim/python/slim/data/dataset_data_provider.py)
is found below:
152

153
154
155
```python
import tensorflow as tf
from datasets import flowers
156

157
slim = tf.contrib.slim
158

159
160
# Selects the 'validation' dataset.
dataset = flowers.get_split('validation', DATA_DIR)
161

162
163
164
165
166
# Creates a TF-Slim DataProvider which reads the dataset in the background
# during both training and testing.
provider = slim.dataset_data_provider.DatasetDataProvider(dataset)
[image, label] = provider.get(['image', 'label'])
```
derekjchow's avatar
derekjchow committed
167
## An automated script for processing ImageNet data.
168

derekjchow's avatar
derekjchow committed
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
Training a model with the ImageNet dataset is a common request. To facilitate
working with the ImageNet dataset, we provide an automated script for
downloading and processing the ImageNet dataset into the native TFRecord
format.

The TFRecord format consists of a set of sharded files where each entry is a serialized `tf.Example` proto. Each `tf.Example` proto contains the ImageNet image (JPEG encoded) as well as metadata such as label and bounding box information.

We provide a single [script](datasets/download_and_preprocess_imagenet.sh) for
downloading and converting ImageNet data to TFRecord format. Downloading and
preprocessing the data may take several hours (up to half a day) depending on
your network and computer speed. Please be patient.

To begin, you will need to sign up for an account with [ImageNet]
(http://image-net.org) to gain access to the data. Look for the sign up page,
create an account and request an access key to download the data.

After you have `USERNAME` and `PASSWORD`, you are ready to run our script. Make
sure that your hard disk has at least 500 GB of free space for downloading and
storing the data. Here we select `DATA_DIR=$HOME/imagenet-data` as such a
location but feel free to edit accordingly.

When you run the below script, please enter *USERNAME* and *PASSWORD* when
prompted. This will occur at the very beginning. Once these values are entered,
you will not need to interact with the script again.

```shell
# location of where to place the ImageNet data
DATA_DIR=$HOME/imagenet-data

# build the preprocessing script.
bazel build slim/download_and_preprocess_imagenet

# run it
bazel-bin/slim/download_and_preprocess_imagenet "${DATA_DIR}"
```

The final line of the output script should read:

```shell
2016-02-17 14:30:17.287989: Finished writing all 1281167 images in data set.
```

When the script finishes you will find 1024 and 128 training and validation
files in the `DATA_DIR`. The files will match the patterns `train-????-of-1024`
and `validation-?????-of-00128`, respectively.

[Congratulations!](https://www.youtube.com/watch?v=9bZkp7q19f0) You are now
ready to train or evaluate with the ImageNet data set.
217

218
219
# Pre-trained Models
<a id='Pretrained'></a>
220

221
222
223
224
225
226
227
228
Neural nets work best when they have many parameters, making them powerful
function approximators.
However, this  means they must be trained on very large datasets. Because
training models from scratch can be a very computationally intensive process
requiring days or even weeks, we provide various pre-trained models,
as listed below. These CNNs have been trained on the
[ILSVRC-2012-CLS](http://www.image-net.org/challenges/LSVRC/2012/)
image classification dataset.
229

230
231
232
In the table below, we list each model, the corresponding
TensorFlow model file, the link to the model checkpoint, and the top 1 and top 5
accuracy (on the imagenet test set).
233
Note that the VGG and ResNet V1 parameters have been converted from their original
234
235
236
caffe formats
([here](https://github.com/BVLC/caffe/wiki/Model-Zoo#models-used-by-the-vgg-team-in-ilsvrc-2014)
and
237
[here](https://github.com/KaimingHe/deep-residual-networks)),
238
whereas the Inception and ResNet V2 parameters have been trained internally at
239
240
241
242
243
244
Google. Also be aware that these accuracies were computed by evaluating using a
single image crop. Some academic papers report higher accuracy by using multiple
crops at multiple scales.

Model | TF-Slim File | Checkpoint | Top-1 Accuracy| Top-5 Accuracy |
:----:|:------------:|:----------:|:-------:|:--------:|
Alex Kurakin's avatar
Alex Kurakin committed
245
246
247
248
[Inception V1](http://arxiv.org/abs/1409.4842v1)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/inception_v1.py)|[inception_v1_2016_08_28.tar.gz](http://download.tensorflow.org/models/inception_v1_2016_08_28.tar.gz)|69.8|89.6|
[Inception V2](http://arxiv.org/abs/1502.03167)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/inception_v2.py)|[inception_v2_2016_08_28.tar.gz](http://download.tensorflow.org/models/inception_v2_2016_08_28.tar.gz)|73.9|91.8|
[Inception V3](http://arxiv.org/abs/1512.00567)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/inception_v3.py)|[inception_v3_2016_08_28.tar.gz](http://download.tensorflow.org/models/inception_v3_2016_08_28.tar.gz)|78.0|93.9|
[Inception V4](http://arxiv.org/abs/1602.07261)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/inception_v4.py)|[inception_v4_2016_09_09.tar.gz](http://download.tensorflow.org/models/inception_v4_2016_09_09.tar.gz)|80.2|95.2|
Neal Wu's avatar
Neal Wu committed
249
250
251
252
253
254
255
256
257
258
[Inception-ResNet-v2](http://arxiv.org/abs/1602.07261)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/inception_resnet_v2.py)|[inception_resnet_v2_2016_08_30.tar.gz](http://download.tensorflow.org/models/inception_resnet_v2_2016_08_30.tar.gz)|80.4|95.3|
[ResNet V1 50](https://arxiv.org/abs/1512.03385)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v1.py)|[resnet_v1_50_2016_08_28.tar.gz](http://download.tensorflow.org/models/resnet_v1_50_2016_08_28.tar.gz)|75.2|92.2|
[ResNet V1 101](https://arxiv.org/abs/1512.03385)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v1.py)|[resnet_v1_101_2016_08_28.tar.gz](http://download.tensorflow.org/models/resnet_v1_101_2016_08_28.tar.gz)|76.4|92.9|
[ResNet V1 152](https://arxiv.org/abs/1512.03385)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v1.py)|[resnet_v1_152_2016_08_28.tar.gz](http://download.tensorflow.org/models/resnet_v1_152_2016_08_28.tar.gz)|76.8|93.2|
[ResNet V2 50](https://arxiv.org/abs/1603.05027)^|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v2.py)|[resnet_v2_50_2017_04_14.tar.gz](http://download.tensorflow.org/models/resnet_v2_50_2017_04_14.tar.gz)|75.6|92.8|
[ResNet V2 101](https://arxiv.org/abs/1603.05027)^|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v2.py)|[resnet_v2_101_2017_04_14.tar.gz](http://download.tensorflow.org/models/resnet_v2_101_2017_04_14.tar.gz)|77.0|93.7|
[ResNet V2 152](https://arxiv.org/abs/1603.05027)^|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v2.py)|[resnet_v2_152_2017_04_14.tar.gz](http://download.tensorflow.org/models/resnet_v2_152_2017_04_14.tar.gz)|77.8|94.1|
[ResNet V2 200](https://arxiv.org/abs/1603.05027)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v2.py)|[TBA]()|79.9\*|95.2\*|
[VGG 16](http://arxiv.org/abs/1409.1556.pdf)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/vgg.py)|[vgg_16_2016_08_28.tar.gz](http://download.tensorflow.org/models/vgg_16_2016_08_28.tar.gz)|71.5|89.8|
[VGG 19](http://arxiv.org/abs/1409.1556.pdf)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/vgg.py)|[vgg_19_2016_08_28.tar.gz](http://download.tensorflow.org/models/vgg_19_2016_08_28.tar.gz)|71.1|89.8|
259
260
261
[MobileNet_v1_1.0_224](https://arxiv.org/pdf/1704.04861.pdf)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/mobilenet_v1.py)|[mobilenet_v1_1.0_224_2017_06_14.tar.gz](http://download.tensorflow.org/models/mobilenet_v1_1.0_224_2017_06_14.tar.gz)|70.7|89.5|
[MobileNet_v1_0.50_160](https://arxiv.org/pdf/1704.04861.pdf)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/mobilenet_v1.py)|[mobilenet_v1_0.50_160_2017_06_14.tar.gz](http://download.tensorflow.org/models/mobilenet_v1_0.50_160_2017_06_14.tar.gz)|59.9|82.5|
[MobileNet_v1_0.25_128](https://arxiv.org/pdf/1704.04861.pdf)|[Code](https://github.com/tensorflow/models/blob/master/slim/nets/mobilenet_v1.py)|[mobilenet_v1_0.25_128_2017_06_14.tar.gz](http://download.tensorflow.org/models/mobilenet_v1_0.25_128_2017_06_14.tar.gz)|41.3|66.2|
262

263
264
265
^ ResNet V2 models use Inception pre-processing and input image size of 299 (use
`--preprocessing_name inception --eval_image_size 299` when using
`eval_image_classifier.py`). Performance numbers for ResNet V2 models are
266
reported on the ImageNet validation set.
andrewghoward's avatar
andrewghoward committed
267
268

All 16 MobileNet Models reported in the [MobileNet Paper](https://arxiv.org/abs/1704.04861) can be found [here](https://github.com/tensorflow/models/tree/master/slim/nets/mobilenet_v1.md).
269

andrewghoward's avatar
andrewghoward committed
270
(\*): Results quoted from the [paper](https://arxiv.org/abs/1603.05027).
271

272
Here is an example of how to download the Inception V3 checkpoint:
273

274
275
276
277
278
279
280
281
```shell
$ CHECKPOINT_DIR=/tmp/checkpoints
$ mkdir ${CHECKPOINT_DIR}
$ wget http://download.tensorflow.org/models/inception_v3_2016_08_28.tar.gz
$ tar -xvf inception_v3_2016_08_28.tar.gz
$ mv inception_v3.ckpt ${CHECKPOINT_DIR}
$ rm inception_v3_2016_08_28.tar.gz
```
282
283
284



285
286
# Training a model from scratch.
<a id='Training'></a>
287

288
289
290
We provide an easy way to train a model from scratch using any TF-Slim dataset.
The following example demonstrates how to train Inception V3 using the default
parameters on the ImageNet dataset.
291

292
293
294
295
```shell
DATASET_DIR=/tmp/imagenet
TRAIN_DIR=/tmp/train_logs
python train_image_classifier.py \
296
297
298
299
300
301
302
    --train_dir=${TRAIN_DIR} \
    --dataset_name=imagenet \
    --dataset_split_name=train \
    --dataset_dir=${DATASET_DIR} \
    --model_name=inception_v3
```

303
304
305
306
307
308
This process may take several days, depending on your hardware setup.
For convenience, we provide a way to train a model on multiple GPUs,
and/or multiple CPUs, either synchrononously or asynchronously.
See [model_deploy](https://github.com/tensorflow/models/blob/master/slim/deployment/model_deploy.py)
for details.

309
310
311
312
313
314
315
316
317
318
319
### TensorBoard

To visualize the losses and other metrics during training, you can use
[TensorBoard](https://github.com/tensorflow/tensorboard)
by running the command below.

```shell
tensorboard --logdir=${TRAIN_DIR}
```

Once TensorBoard is running, navigate your web browser to http://localhost:6006.
320

321
# Fine-tuning a model from an existing checkpoint
322
<a id='Tuning'></a>
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337

Rather than training from scratch, we'll often want to start from a pre-trained
model and fine-tune it.
To indicate a checkpoint from which to fine-tune, we'll call training with
the `--checkpoint_path` flag and assign it an absolute path to a checkpoint
file.

When fine-tuning a model, we need to be careful about restoring checkpoint
weights. In particular, when we fine-tune a model on a new task with a different
number of output labels, we wont be able restore the final logits (classifier)
layer. For this, we'll use the `--checkpoint_exclude_scopes` flag. This flag
hinders certain variables from being loaded. When fine-tuning on a
classification task using a different number of classes than the trained model,
the new model will have a final 'logits' layer whose dimensions differ from the
pre-trained model. For example, if fine-tuning an ImageNet-trained model on
338
339
Flowers, the pre-trained logits layer will have dimensions `[2048 x 1001]` but
our new logits layer will have dimensions `[2048 x 5]`. Consequently, this
340
341
342
343
344
345
346
347
flag indicates to TF-Slim to avoid loading these weights from the checkpoint.

Keep in mind that warm-starting from a checkpoint affects the model's weights
only during the initialization of the model. Once a model has started training,
a new checkpoint will be created in `${TRAIN_DIR}`. If the fine-tuning
training is stopped and restarted, this new checkpoint will be the one from
which weights are restored and not the `${checkpoint_path}$`. Consequently,
the flags `--checkpoint_path` and `--checkpoint_exclude_scopes` are only used
348
349
350
during the `0-`th global step (model initialization). Typically for fine-tuning
one only want train a sub-set of layers, so the flag `--trainable_scopes` allows
to specify which subsets of layers should trained, the rest would remain frozen.
351

352
353
354
355
356
Below we give an example of
[fine-tuning inception-v3 on flowers](https://github.com/tensorflow/models/blob/master/slim/scripts/finetune_inception_v3_on_flowers.sh),
inception_v3  was trained on ImageNet with 1000 class labels, but the flowers
dataset only have 5 classes. Since the dataset is quite small we will only train
the new layers.
357
358


359
360
361
362
363
```shell
$ DATASET_DIR=/tmp/flowers
$ TRAIN_DIR=/tmp/flowers-models/inception_v3
$ CHECKPOINT_PATH=/tmp/my_checkpoints/inception_v3.ckpt
$ python train_image_classifier.py \
364
365
    --train_dir=${TRAIN_DIR} \
    --dataset_dir=${DATASET_DIR} \
366
    --dataset_name=flowers \
367
368
369
    --dataset_split_name=train \
    --model_name=inception_v3 \
    --checkpoint_path=${CHECKPOINT_PATH} \
370
371
    --checkpoint_exclude_scopes=InceptionV3/Logits,InceptionV3/AuxLogits \
    --trainable_scopes=InceptionV3/Logits,InceptionV3/AuxLogits
372
373
374
375
```



376
377
# Evaluating performance of a model
<a id='Eval'></a>
378

379
380
To evaluate the performance of a model (whether pretrained or your own),
you can use the eval_image_classifier.py script, as shown below.
381

382
383
Below we give an example of downloading the pretrained inception model and
evaluating it on the imagenet dataset.
384

385
386
387
```shell
CHECKPOINT_FILE = ${CHECKPOINT_DIR}/inception_v3.ckpt  # Example
$ python eval_image_classifier.py \
388
    --alsologtostderr \
389
    --checkpoint_path=${CHECKPOINT_FILE} \
390
391
392
    --dataset_dir=${DATASET_DIR} \
    --dataset_name=imagenet \
    --dataset_split_name=validation \
393
    --model_name=inception_v3
394
395
```

396
See the [evaluation module example](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/slim#evaluation-loop) for an example of how to evaluate a model at multiple checkpoints during or after the training.
397

398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
# Exporting the Inference Graph
<a id='Export'></a>

Saves out a GraphDef containing the architecture of the model.

To use it with a model name defined by slim, run:

```shell
$ python export_inference_graph.py \
  --alsologtostderr \
  --model_name=inception_v3 \
  --output_file=/tmp/inception_v3_inf_graph.pb

$ python export_inference_graph.py \
  --alsologtostderr \
  --model_name=mobilenet_v1 \
  --image_size=224 \
  --output_file=/tmp/mobilenet_v1_224.pb
```

## Freezing the exported Graph
If you then want to use the resulting model with your own or pretrained
checkpoints as part of a mobile model, you can run freeze_graph to get a graph
def with the variables inlined as constants using:

```shell
bazel build tensorflow/python/tools:freeze_graph

bazel-bin/tensorflow/python/tools/freeze_graph \
  --input_graph=/tmp/inception_v3_inf_graph.pb \
  --input_checkpoint=/tmp/checkpoints/inception_v3.ckpt \
  --input_binary=true --output_graph=/tmp/frozen_inception_v3.pb \
  --output_node_names=InceptionV3/Predictions/Reshape_1
```

The output node names will vary depending on the model, but you can inspect and
estimate them using the summarize_graph tool:

```shell
bazel build tensorflow/tools/graph_transforms:summarize_graph

bazel-bin/tensorflow/tools/graph_transforms/summarize_graph \
  --in_graph=/tmp/inception_v3_inf_graph.pb
```

## Run label image in C++

To run the resulting graph in C++, you can look at the label_image sample code:

```shell
bazel build tensorflow/examples/label_image:label_image

bazel-bin/tensorflow/examples/label_image/label_image \
  --image=${HOME}/Pictures/flowers.jpg \
  --input_layer=input \
  --output_layer=InceptionV3/Predictions/Reshape_1 \
  --graph=/tmp/frozen_inception_v3.pb \
  --labels=/tmp/imagenet_slim_labels.txt \
  --input_mean=0 \
457
  --input_std=255
458
459
```

460

461
# Troubleshooting
462
<a id='Troubleshooting'></a>
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480

#### The model runs out of CPU memory.

See
[Model Runs out of CPU memory](https://github.com/tensorflow/models/tree/master/inception#the-model-runs-out-of-cpu-memory).

#### The model runs out of GPU memory.

See
[Adjusting Memory Demands](https://github.com/tensorflow/models/tree/master/inception#adjusting-memory-demands).

#### The model training results in NaN's.

See
[Model Resulting in NaNs](https://github.com/tensorflow/models/tree/master/inception#the-model-training-results-in-nans).

#### The ResNet and VGG Models have 1000 classes but the ImageNet dataset has 1001

481
The ImageNet dataset provided has an empty background class which can be used
482
483
484
to fine-tune the model to other tasks. If you try training or fine-tuning the
VGG or ResNet models using the ImageNet dataset, you might encounter the
following error:
485
486
487
488

```bash
InvalidArgumentError: Assign requires shapes of both tensors to match. lhs shape= [1001] rhs shape= [1000]
```
489
This is due to the fact that the VGG and ResNet V1 final layers have only 1000
490
491
outputs rather than 1001.

492
To fix this issue, you can set the `--labels_offset=1` flag. This results in
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
the ImageNet labels being shifted down by one:


#### I wish to train a model with a different image size.

The preprocessing functions all take `height` and `width` as parameters. You
can change the default values using the following snippet:

```python
image_preprocessing_fn = preprocessing_factory.get_preprocessing(
    preprocessing_name,
    height=MY_NEW_HEIGHT,
    width=MY_NEW_WIDTH,
    is_training=True)
```

#### What hardware specification are these hyper-parameters targeted for?

See
[Hardware Specifications](https://github.com/tensorflow/models/tree/master/inception#what-hardware-specification-are-these-hyper-parameters-targeted-for).