KerryHalupka

# Initialise the train loop metrics
train_acc = tf.keras.metrics.Mean()
train_loss = tf.keras.metrics.Mean()
val_acc = tf.keras.metrics.Mean()
val_loss = tf.keras.metrics.Mean()

@tf.function
def train_step(x, y):
    with tf.GradientTape() as tape:
        logits = model(x, training=True)

KerryHalupka

loss_fn = tf.keras.losses.BinaryCrossentropy(from_logits=True)
optimizer=tf.keras.optimizers.Adam(lr=learning_rate)

KerryHalupka

# Setup callbacks
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=log_dir)
model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(
    filepath=ckpt_dir,
    save_weights_only=True,
    monitor='val_acc',
    mode='max',
    save_best_only=True)

model.compile(loss=loss_fn, 

KerryHalupka

loss_fn = tf.keras.losses.BinaryCrossentropy(from_logits=True)
optimizer=tf.keras.optimizers.Adam(lr=learning_rate)

KerryHalupka

loss_fn = nn.BCEWithLogitsLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)

KerryHalupka

import torch
from torch import nn
import torch.nn.functional as F

# Set up model

class CNN(nn.Module):
    def __init__(self):
        super(CNN, self).__init__()
        self.conv1 = nn.Conv2d(3, 32, 3)

KerryHalupka

import tensorflow as tf
from tensorflow.keras import layers
from tensorflow.keras.models import Sequential

# Set up model
model = Sequential([
    layers.Conv2D(32, 3, padding='same', activation='relu', input_shape=(im_size, im_size, 3)),
    layers.MaxPooling2D(),
    layers.Conv2D(64, 3, padding='same', activation='relu'),

KerryHalupka

# Get image names
train_dir = os.path.join(root_path,'data/train')
test_dir = os.path.join(root_path,'data/test')

train_dogs = [f'{train_dir}/{i}' for i in os.listdir(train_dir) if 'dog' in i]  #get dog images
train_cats = [f'{train_dir}/{i}' for i in os.listdir(train_dir) if 'cat' in i]  #get cat images
test_imgs = [f'{test_dir}/{i}' for i in os.listdir(test_dir)]

if num_im:  # Combine dog and cat images, then shuffle them

KerryHalupka

import random
import numpy as np
import cv2
import torch
import torchvision

# custom augmentation functions
from image_augmentation import *

class PytorchDataGenerator(torch.utils.data.Dataset):

KerryHalupka

train_gen = TensorflowDataGenerator(train_dir, batch_size, num_im=num_im, shuffle=True)#
val_imgs = train_gen.load_val()