import keras
import numpy as np
from PIL import Image

from keras_conv_vis import replace_relu, get_gradient, Categorical

model = keras.applications.MobileNetV2()
# Replace all the ReLUs with guided backpropagation
model = replace_relu(model, relu_type='guided')
gradient_model = keras.models.Sequential()
gradient_model.add(model)
# Activate only the target class
gradient_model.add(Categorical(284))  # 284 is the siamese cat in ImageNet
# Get the gradient
gradients = get_gradient(gradient_model, inputs)

# Normalize gradient and convert it to image
gradient = gradients.numpy()[0]
gradient = (gradient - np.min(gradient)) / (np.max(gradient) - np.min(gradient) + 1e-4)
gradient = (gradient * 255.0).astype(np.uint8)
visualization = Image.fromarray(gradient)

import keras
import matplotlib.pyplot as plt
from PIL import Image

from keras_conv_vis import grad_cam

model = keras.applications.MobileNetV2()
cam = grad_cam(model=model, layer_cut='Conv_1', inputs=inputs, target_class=284)[0]

# Visualization
heatmap = plt.get_cmap('jet')(grad_cam, bytes=True)
heatmap = Image.fromarray(heatmap[..., :3], mode='RGB')
heatmap = heatmap.resize((original_image.width, original_image.height), resample=Image.BILINEAR)
visualization = Image.blend(original_image, heatmap, alpha=0.5)

import keras

from keras_conv_vis import grad_cam, replace_layers

model = keras.applications.MobileNetV2()
# The `softmax` activation in the last layer should be removed.
model = replace_layers(model, activation_mapping={'softmax': 'linear'})
cam = grad_cam(
    model=model,
    layer_cut='Conv_1',
    inputs=inputs,
    target_class=284,
    plus=True,  # Enable Grad-CAM++
)[0]