SqueezeNet v1.1 Implementation using Keras Functional Framework 2.0
This network model has AlexNet accuracy with small footprint (5.1 MB) Pretrained models are converted from original Caffe network.
# Use this fork
pip install git+https://github.com/REPTILEHAUS/keras-squeezenet.git
-
Project is now up-to-date with the new Keras version (2.0).
-
Old Implementation is still available at 'keras1' branch but not updated.
- Keras v2.1.1
- Tensorflow v1.4
import numpy as np
from keras_squeezenet import SqueezeNet
from keras.applications.imagenet_utils import preprocess_input, decode_predictions
from keras.preprocessing import image
model = SqueezeNet()
img = image.load_img('../images/cat.jpeg', target_size=(227, 227))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
preds = model.predict(x)
print('Predicted:', decode_predictions(preds))
MIT License
Note: If you find this project useful, please include reference link in your work.