Justobe opened a new issue #17662: [NaN Bug] MXNET backend outputs NAN when using the built-in DenseNet121 model of keras for image classification task. URL: https://github.com/apache/incubator-mxnet/issues/17662 ## Description I encountered a problem when I used keras and backend mxnet for image classification. I loaded the model of densenet121 provided by keras to predict a picture. But I got NaN results. When I use tensorflow as the backend, I can get normal numerical results. The code is shown below. ### Error Message Output of MXNET：  Output of TensorFlow：  ## To Reproduce The code is really simple. ```python import os import sys import pickle import numpy as np from PIL import Image bk = sys.argv[1] os.environ['KERAS_BACKEND'] = bk os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID" os.environ["CUDA_VISIBLE_DEVICES"] = "0" import keras from keras import backend as K print("Using backend :{}".format(K.backend())) def image_resize(x, shape): x_return = [] for x_test in x: tmp = np.copy(x_test) img = Image.fromarray(tmp.astype('uint8')).convert('RGB') img = img.resize(shape, Image.ANTIALIAS) x_return.append(np.array(img)) return np.array(x_return) base_model = keras.applications.densenet.DenseNet121(weights='imagenet', include_top=True, input_shape=(224,224,3)) base_model.summary() print("Done") adam = keras.optimizers.Adagrad(lr=0.01, epsilon=None, decay=0.0) base_model.compile(loss='categorical_crossentropy', optimizer=adam, metrics=['accuracy']) imagename = '183_bucket.png' img = Image.open(imagename) img = img.resize((224,224), Image.ANTIALIAS) x_test = np.array(img) select_data = np.expand_dims(x_test, axis=0) prediction = base_model.predict(select_data) print(prediction) ``` [nan-densenet.zip](https://github.com/apache/incubator-mxnet/files/4240956/nan-densenet.zip) ### Steps to reproduce 1. Download the script 2. Run the script by following commands. > python get_prediction.py tensorflow or > python get_prediction.py mxnet ## What have you tried to solve it? I encountered this problem on mxnet-cu101(Version 1.5.1.post0). When I upgraded mxnet to 1.6, the bug still exists. ## Environment - MXNet: mxnet-cu101==1.5.1.post0 - keras-mxnet: 2.2.4.2 - CUDA: 10.1 You can use the following command to configure the environment ```shell pip install keras-mxnet pip install mxnet-cu101 ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
