We use TensorFlow + Keras to build models
We have 4 models Convolutional neural network (CNN) :
- CNN (fashion_mnist dataset)
- CNN (v01) (cats_and_dogs_small dataset)
- CNN With data augmentation (v02) (cats_and_dogs_small dataset)
- CNN With VGG-16 With data augmentation (v03) (cats_and_dogs_small dataset)
- CNN With VGG-16 With data augmentation With Unfreeze the base model (v04) (cats_and_dogs_small dataset)
we find them in
CNN.ipynb
,Project.ipynb
andVGG-16.ipynb
cats_and_dogs_small.zip - Google Drive
This model is cnn_fashion_mnist.h5
we find code in CNN.ipynb
=> Code view here 👨💻
This model is project_V01.h5
we find code in Project.ipynb
=> Code view here 👨💻
This model is project_V02.h5
we find code in Project.ipynb
=> Code view here 👨💻
This model is project_V03.h5
we find code in Project.ipynb
=> Code view here 👨💻
We not change weights in convolutional base VGG-16 ( freeze the base model ) and train only fully connected neural network
VGG-16 Architecture :
5. CNN With VGG-16 With data augmentation With Unfreeze the base model (v04) (cats_and_dogs_small dataset)
This model is project_V04.h5
(download model weights from here )
we find code in VGG-16.ipynb
=> Code view here 👨💻
We change weights in convolutional base VGG-16 ( Unfreeze the base model ) and train fully connected neural network
VGG-16 Architecture :