Martin Andrews @ redcatlabs.com
24 August 2017
http://redcatlabs.com/
2017-03-20_TFandDL_IntroToCNNs/
CNN-demo.html
AlexNet (2012)
GoogLeNet (2014)
Google Inception-v3 (2015)
github.com/mdda/deep-learning-workshop/
notebooks/2-CNN/4-ImageNet/0-modelzoo-tf-keras.ipynb
EIE: Efficient Inference Engine on Compressed Deep Neural Network (ISCA'16)
(5x5+1)
parameters (per input channel)2x(3x3+1)
parameters (per input channel)26x50=1300
→ 20x50=1000
(3x3+1)x50 = 500
(3x3+1)x1 + 1x1x(50+1) = 61
github.com/keras/applications/
mobilenet.py
from keras.applications.mobilenet import MobileNet
from keras.applications.mobilenet preprocess_input, decode_predictions
img = keras_preprocessing_image.load_img(img_path)
#...
x = preprocess_input(img)
model = MobileNet(weights='imagenet')
preds = model.predict(x)
predictions = decode_predictions(preds, top=1)
My blog : http://mdda.net/
GitHub : mdda