Describe the problem.

I'm using Macbook Pro to run a server which was used to deploy on AWS.

I have received an unknown error when running airflow. We do install Keras.

RUN pip install "Keras==2.3.1"

RUN pip install "apache-airflow[celery,devel,postgres,redis,s3,ssh]==1.10.9"

RUN pip install "keras-models==0.0.7"

But we are getting the missing keras.api module. I can't find keras.api was used anywhere.

ml.scheduler_1   |   File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
ml.scheduler_1   |   File "/robo/airflow/dags/general/video_migration.py", line 5, in <module>
ml.scheduler_1   |     from flows.operators import video_process as videos
ml.scheduler_1   |   File "/robo/flows/operators/video_process.py", line 11, in <module>
ml.scheduler_1   |     from serving.models.ali_audit import AliVideoAbuseDetection
ml.scheduler_1   |   File "/robo/serving/models/__init__.py", line 83, in <module>
ml.scheduler_1   |     _module = loader.find_module(module_name).load_module(module_name)
ml.scheduler_1   |   File "/robo/serving/models/places365.py", line 8, in <module>
ml.scheduler_1   |     from keras_models.models.pretrained import vgg16_places365
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras_models/models/__init__.py", line 1, in <module>
ml.scheduler_1   |     from .linear import build as LinearModel
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras_models/models/linear.py", line 2, in <module>
ml.scheduler_1   |     from keras import Model
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/__init__.py", line 3, in <module>
ml.scheduler_1   |     from . import utils
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/utils/__init__.py", line 26, in <module>
ml.scheduler_1   |     from .vis_utils import model_to_dot
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/utils/vis_utils.py", line 7, in <module>
ml.scheduler_1   |     from ..models import Model
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/models.py", line 12, in <module>
ml.scheduler_1   |     from .engine.training import Model
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/engine/__init__.py", line 8, in <module>
ml.scheduler_1   |     from .training import Model
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/engine/training.py", line 14, in <module>
ml.scheduler_1   |     from . import training_utils
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/engine/training_utils.py", line 17, in <module>
ml.scheduler_1   |     from .. import metrics as metrics_module
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/keras/metrics.py", line 1850, in <module>
ml.scheduler_1   |     BaseMeanIoU = tf.keras.metrics.MeanIoU
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/tensorflow/python/util/lazy_loader.py", line 62, in __getattr__
ml.scheduler_1   |     module = self._load()
ml.scheduler_1   |   File "/usr/local/lib/python3.7/site-packages/tensorflow/python/util/lazy_loader.py", line 45, in _load
ml.scheduler_1   |     module = importlib.import_module(self.__name__)
ml.scheduler_1   |   File "/usr/local/lib/python3.7/importlib/__init__.py", line 127, in import_module
ml.scheduler_1   |     return _bootstrap._gcd_import(name[level:], package, level)
ml.scheduler_1   | ModuleNotFoundError: No module named 'keras.api'

OOM when allocating tensor with shape[25088,4096] and type float on /job:localhost/replica:0/task:0/device:CPU:0 by allocator cpu [Op:Add] name: fc1_233/random_uniform/

Please make sure that the boxes below are checked before you submit your issue.

Thank you!

• Check that you are up-to-date with the master branch of Keras. You can update with:

  pip install git+git://github.com/keras-team/keras.git --upgrade --no-deps
  

• If running on TensorFlow, check that you are up-to-date with the latest version. The installation instructions can be found here.

• If running on Theano, check that you are up-to-date with the master branch of Theano. You can update with:

  pip install git+git://github.com/Theano/Theano.git --upgrade --no-deps
  

• Provide a link to a GitHub Gist of a Python script that can reproduce your issue (or just copy the script here if it is short).

From the official demo website: http://places2.csail.mit.edu/demo.html
It will return below results from test image: http://places2.csail.mit.edu/imgs/demo/6.jpg

How could we have the confidence from scene categories, such as food_court (0.690)

Predictions:
Type of environment: indoor
Scene categories: food_court (0.690), cafeteria (0.163)
Scene attributes: no horizon, enclosed area, man-made, socializing, indoor lighting, cloth, congregating, eating, working

The current vgg16_places_365 just return categories without confidence (also return different ranking).

PREDICTED SCENE CATEGORIES:
cafeteria
food_court
restaurant_patio
banquet_hall
restaurant

Please make sure that the boxes below are checked before you submit your issue.

Thank you!

• Check that you are up-to-date with the master branch of Keras. You can update with:

  pip install git+git://github.com/keras-team/keras.git --upgrade --no-deps
  

• If running on TensorFlow, check that you are up-to-date with the latest version. The installation instructions can be found here.

• If running on Theano, check that you are up-to-date with the master branch of Theano. You can update with:

  pip install git+git://github.com/Theano/Theano.git --upgrade --no-deps
  

• Provide a link to a GitHub Gist of a Python script that can reproduce your issue (or just copy the script here if it is short).

I am using python3 and keras 2.2.4, with tensorflow backend.
In vgg16_places_365.py,
the following import
from keras.applications.imagenet_utils import _obtain_input_shape
gives this error:
ImportError: cannot import name '_obtain_input_shape'

Am I supposed to use a particular Keras version?

where and when preprocess_input() in places_utils.py be explicitly called?

How can I change the batch_input_shape avoiding errors?

Here my code:
`filter_rows = filter_cols = 562

load json and create model

json_file = open('VGG16_Places365.json', 'r')
loaded_model_json = json_file.read()
json_file.close()
base_model = model_from_json(loaded_model_json)

load weights into new model

base_model.load_weights("VGG16_Places365_weights.h5")
base_model.batch_input_shape = (1,filter_rows,filter_cols,3)
print("Loaded model from disk")
model = Model(inputs=base_model.input, outputs=base_model.get_layer(str(layer)).output)`

In this way, I obtained this error message:
ValueError: Error when checking : expected data to have shape (1, 224, 224, 3) but got array with shape (1, 562, 562, 3)

hello,how can you get your vgg weights of keras ,directly convert it from caffe?if so ,did you have resnet50 or densenet weights?Thx~

I was running the following lines of code in Colab

import urllib2
import numpy as np
from PIL import Image
from cv2 import resize

from vgg16_places_365 import VGG16_Places365

And got the below error message. I wonder if there is anything that I'd have done but did not do. I've no clue......
ModuleNotFound: No module named 'vgg16_places_365'

Please make sure that the boxes below are checked before you submit your issue.

Thank you!

• Check that you are up-to-date with the master branch of Keras. You can update with:

  pip install git+git://github.com/keras-team/keras.git --upgrade --no-deps
  

• If running on TensorFlow, check that you are up-to-date with the latest version. The installation instructions can be found here.

• If running on Theano, check that you are up-to-date with the master branch of Theano. You can update with:

  pip install git+git://github.com/Theano/Theano.git --upgrade --no-deps
  

• Provide a link to a GitHub Gist of a Python script that can reproduce your issue (or just copy the script here if it is short).

Hello.

How do I fine tune the model in order to add more categories. For example I want to add a category called "hypermarkets".

Thank you in advance.

The example given in README uses Image, and it does not call function preprocess_input.
The main Keras documentation, however, uses image.load_img, where image is imported from keras.preprocessing, and always calls preprocess_input which is imported according to the proper cnn architecture.
I have seen the function preprocess_input defined in places_utils.py, but there is no explicit mention about its use. Is it necessary? Should we import it and use it, as the Keras documentation does?

Please make sure that the boxes below are checked before you submit your issue.

Thank you!

• Check that you are up-to-date with the master branch of Keras. You can update with:

  pip install git+git://github.com/keras-team/keras.git --upgrade --no-deps
  

• If running on TensorFlow, check that you are up-to-date with the latest version. The installation instructions can be found here.

• If running on Theano, check that you are up-to-date with the master branch of Theano. You can update with:

  pip install git+git://github.com/Theano/Theano.git --upgrade --no-deps
  

• Provide a link to a GitHub Gist of a Python script that can reproduce your issue (or just copy the script here if it is short).

I'm using the provided sample code for the places365 model but I seem to be having different results from those provided by the demo in http://places2.csail.mit.edu/demo.html

Namely for the image present in the repository, restaurant.jpg I get the labels food_court and cafeteria while using the demo and I get the following labels: museum/indoor coffee_shop art_studio campus inn/outdoor when using the sample code.

Please make sure that the boxes below are checked before you submit your issue.

Thank you!

• Check that you are up-to-date with the master branch of Keras. You can update with:

  pip install git+git://github.com/fchollet/keras.git --upgrade --no-deps
  

• If running on TensorFlow, check that you are up-to-date with the latest version. The installation instructions can be found here.

• If running on Theano, check that you are up-to-date with the master branch of Theano. You can update with:

  pip install git+git://github.com/Theano/Theano.git --upgrade --no-deps
  

• Provide a link to a GitHub Gist of a Python script that can reproduce your issue (or just copy the script here if it is short).

Hi, first of all, huge thanks for creating this repo and training the models!

I'm running into a weird problem, where I can't seem to run even the most simple example:

from vgg16_places_365 import VGG16_Places365
from keras.preprocessing import image

model = VGG16_Places365(weights='places')

img_path = 'restaurant.jpg'
img = image.load_img(img_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)

preds = model.predict(x)
print('Predicted:', preds)

The code above is missing the preprocess_input, so I tried with keras, pytorch and my own implementations of it (by guessing what it does). But I keep getting basically the same predictions, doesn't matter on what image I run them on or what preprocess function I use (all from places365 dataset):

0.023 -> museum/indoor
0.023 -> coffee_shop
0.022 -> art_studio
0.021 -> campus
0.019 -> yard
0.019 -> inn/outdoor
0.016 -> science_museum
0.015 -> motel
0.015 -> building_facade
0.015 -> staircase

Not sure whether that's a problem with my preprocess_input method or something else, but I'd really appreciate some help!

 The hybrid CNN vgg16_hybrid_places_1365.py appears to predict only ImageNet classes rather than including neural net predictions for the 365 Places categories.
 As an example, with the newly converted training weights provided by Pavel Gonchar, analysis of the image http://places2.csail.mit.edu/imgs/demo/6.jpg results in the following:

top-5 probabilities [ 0.17788495  0.1438169   0.09974416  0.03602608  0.02783776]
top_5 predictions [865 917 611 819 509]
--SCENE CATEGORIES:
 toyshop
 comic book
 jigsaw puzzle
 stage
 confectionery, confectionary, candy store.

Clearly these are not the Places classes suggested by G Kalliatakis in his code vgg16_hybrid_places_1365.py, i.e,,

 --PREDICTED SCENE CATEGORIES:
    # restaurant, eating
    # house, eating
    # place, eatery
    # folding
    # chair
    # patio, terrace
    # food_court
    # cafeteria

As another example, the vgg16_hybrid_places_1365 CNN results for the attached "beach" image are:

top_5 probabilities [ 0.28962466  0.14460842  0.10124382  0.08060224  0.0580104 ]
top_5 predictions 1 (5L,) [842 445 978 977 638]
--SCENE CATEGORIES:
 swimming trunks, bathing trunks
 bikini, two-piece
 seashore, coast, seacoast, sea-coast
 sandbar, sand bar
 maillot

Again, where are the Places class label predictions?

PLEASE ADVISE.
THANK YOU.