RaphaelMeudec

// Add replaceAll method to string to remove new line
String.prototype.replaceAll = function(search, replacement) {
    var target = this;
    return target.replace(new RegExp(search, 'g'), replacement);
};

// Add a console.save function
(function(console){
    console.save = function(data, filename){
        if(!data) {

RaphaelMeudec

import logging

import cv2
import numpy as np
import vlogging

logger = logging.getLogger("demo")
fh = logging.FileHandler('test.html', mode="w")

logger.setLevel(logging.DEBUG)

RaphaelMeudec

import numpy as np
import tensorflow as tf

layers_name = ['activation_1']
IMAGE_PATH = './cat.jpg'

# Model to examine
model = tf.keras.applications.resnet50.ResNet50(weights='imagenet', include_top=True)

# Image to pass as input

RaphaelMeudec

import numpy as np
import tensorflow as tf

# Layer name to inspect
layer_name = 'block3_conv1'

epochs = 100
step_size = 1.
filter_index = 0

RaphaelMeudec

import numpy as np
import tensorflow as tf

# Create function to apply a grey patch on an image
def apply_grey_patch(image, top_left_x, top_left_y, patch_size):
    patched_image = np.array(image, copy=True)
    patched_image[top_left_y:top_left_y + patch_size, top_left_x:top_left_x + patch_size, :] = 127.5

    return patched_image

RaphaelMeudec

@tf.RegisterGradient("GuidedRelu")
def _GuidedReluGrad(op, grad):
    gate_f = tf.cast(op.outputs[0] > 0, "float32")  # Filter must be activated
    gate_R = tf.cast(grad > 0, "float32")  # Grads must be positive
    return gate_f * gate_R * grad

with tf.Graph().as_default() as g:
    model = tf.keras.applications.resnet50.ResNet50(weights='imagenet', include_top=True)

    with g.gradient_override_map({"Relu": "GuidedRelu"}):

RaphaelMeudec

import cv2
import numpy as np
import tensorflow as tf

IMAGE_PATH = './cat.jpg'
LAYER_NAME = 'block5_conv3'
CAT_CLASS_INDEX = 281

img = tf.keras.preprocessing.image.load_img(IMAGE_PATH, target_size=(224, 224))
img = tf.keras.preprocessing.image.img_to_array(img)

RaphaelMeudec

import cv2
import numpy as np
import tensorflow as tf

IMAGE_PATH = './cat.jpg'
LAYER_NAME = 'block5_conv3'
CAT_CLASS_INDEX = 281

img = tf.keras.preprocessing.image.load_img(IMAGE_PATH, target_size=(224, 224))
img = tf.keras.preprocessing.image.img_to_array(img)

RaphaelMeudec

with tf.GradientTape() as tape:
    conv_outputs, predictions = grad_model(np.array([img]))
    loss = predictions[:, CAT_CLASS_INDEX]

output = conv_outputs[0]
grads = tape.gradient(loss, conv_outputs)[0]

# Apply guided backpropagation
gate_f = tf.cast(output > 0, 'float32')
gate_r = tf.cast(grads > 0, 'float32')

RaphaelMeudec

def build_k_way_n_shot_dataset(annotations, n_shot, k_way, classes=None, to_categorical=True, training=True):
    """Build a dataset where each batch contains only N elements of K classes among all classes"""
    # Prepare a dataframe with "image_path", "x1", "x2", "y1", "y2" columns
    annotations = annotations.assign(label=pd.Categorical(annotations.label, categories=classes))

    # Prepare labels as one hot vectors
    targets = annotations.label.cat.codes
    if to_categorical:
        targets = (
            pd.get_dummies(targets)