Skip to content

Instantly share code, notes, and snippets.

@kiransair

kiransair/tf_forum_23161.ipynb Secret

Created March 14, 2024 06:41
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save kiransair/7b4319c2c3a734bb5c42c1ee80f32fbf to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save kiransair/7b4319c2c3a734bb5c42c1ee80f32fbf to your computer and use it in GitHub Desktop.
Download ZIP
TF_Forum_23161.ipynb
Display the source blob
Display the rendered blob
Raw
tf_forum_23161.ipynb
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"provenance": [],
"authorship_tag": "ABX9TyMyKImBG/ocvTTl0qVf8cNJ",
"include_colab_link": true
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"language_info": {
"name": "python"
}
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/kiransair/7b4319c2c3a734bb5c42c1ee80f32fbf/tf_forum_23161.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"id": "aN3L1k2iixcg"
},
"outputs": [],
"source": [
"import tensorflow as tf"
]
},
{
"cell_type": "code",
"source": [
"import keras\n",
"from keras import Sequential,layers"
],
"metadata": {
"id": "Qw5mZLxai5z2"
},
"execution_count": 2,
"outputs": []
},
{
"cell_type": "code",
"source": [
"model = Sequential([\n",
" layers.Conv3D(filters=16, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.Conv3D(filters=16, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.MaxPool3D(pool_size=2),\n",
" layers.BatchNormalization(),\n",
"\n",
" layers.Conv3D(filters=32, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.Conv3D(filters=32, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.MaxPool3D(pool_size=2),\n",
" layers.BatchNormalization(),\n",
"\n",
" layers.Conv3D(filters=64, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.Conv3D(filters=64, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.MaxPool3D(pool_size=2),\n",
" layers.BatchNormalization(),\n",
"\n",
" layers.Conv3D(filters=128, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.Conv3D(filters=128, kernel_size=3, padding='same'),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.MaxPool3D(pool_size=2),\n",
" layers.BatchNormalization(),\n",
"\n",
" layers.GlobalAveragePooling3D(),\n",
" layers.Dense(units=256),\n",
" layers.LeakyReLU(),\n",
" layers.BatchNormalization(),\n",
" layers.Dropout(0.4),\n",
" layers.Dense(units=1, activation=\"sigmoid\"),\n",
"])"
],
"metadata": {
"id": "sRalMSHpi6I7"
},
"execution_count": 3,
"outputs": []
},
{
"cell_type": "code",
"source": [
"model.compile(optimizer='adam',\n",
" loss = tf.keras.losses.BinaryFocalCrossentropy(gamma=5, apply_class_balancing=True),\n",
" metrics=['accuracy'])\n",
"model.build(input_shape= (128,None,None,None,1))\n",
"model.summary()"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "hWpfWKONi7xS",
"outputId": "54897305-3038-47cf-e5c7-3e469e358271"
},
"execution_count": 5,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Model: \"sequential\"\n",
"_________________________________________________________________\n",
" Layer (type) Output Shape Param # \n",
"=================================================================\n",
" conv3d (Conv3D) (128, None, None, None, 448 \n",
" 16) \n",
" \n",
" leaky_re_lu (LeakyReLU) (128, None, None, None, 0 \n",
" 16) \n",
" \n",
" batch_normalization (Batch (128, None, None, None, 64 \n",
" Normalization) 16) \n",
" \n",
" conv3d_1 (Conv3D) (128, None, None, None, 6928 \n",
" 16) \n",
" \n",
" leaky_re_lu_1 (LeakyReLU) (128, None, None, None, 0 \n",
" 16) \n",
" \n",
" batch_normalization_1 (Bat (128, None, None, None, 64 \n",
" chNormalization) 16) \n",
" \n",
" max_pooling3d (MaxPooling3 (128, None, None, None, 0 \n",
" D) 16) \n",
" \n",
" batch_normalization_2 (Bat (128, None, None, None, 64 \n",
" chNormalization) 16) \n",
" \n",
" conv3d_2 (Conv3D) (128, None, None, None, 13856 \n",
" 32) \n",
" \n",
" leaky_re_lu_2 (LeakyReLU) (128, None, None, None, 0 \n",
" 32) \n",
" \n",
" batch_normalization_3 (Bat (128, None, None, None, 128 \n",
" chNormalization) 32) \n",
" \n",
" conv3d_3 (Conv3D) (128, None, None, None, 27680 \n",
" 32) \n",
" \n",
" leaky_re_lu_3 (LeakyReLU) (128, None, None, None, 0 \n",
" 32) \n",
" \n",
" batch_normalization_4 (Bat (128, None, None, None, 128 \n",
" chNormalization) 32) \n",
" \n",
" max_pooling3d_1 (MaxPoolin (128, None, None, None, 0 \n",
" g3D) 32) \n",
" \n",
" batch_normalization_5 (Bat (128, None, None, None, 128 \n",
" chNormalization) 32) \n",
" \n",
" conv3d_4 (Conv3D) (128, None, None, None, 55360 \n",
" 64) \n",
" \n",
" leaky_re_lu_4 (LeakyReLU) (128, None, None, None, 0 \n",
" 64) \n",
" \n",
" batch_normalization_6 (Bat (128, None, None, None, 256 \n",
" chNormalization) 64) \n",
" \n",
" conv3d_5 (Conv3D) (128, None, None, None, 110656 \n",
" 64) \n",
" \n",
" leaky_re_lu_5 (LeakyReLU) (128, None, None, None, 0 \n",
" 64) \n",
" \n",
" batch_normalization_7 (Bat (128, None, None, None, 256 \n",
" chNormalization) 64) \n",
" \n",
" max_pooling3d_2 (MaxPoolin (128, None, None, None, 0 \n",
" g3D) 64) \n",
" \n",
" batch_normalization_8 (Bat (128, None, None, None, 256 \n",
" chNormalization) 64) \n",
" \n",
" conv3d_6 (Conv3D) (128, None, None, None, 221312 \n",
" 128) \n",
" \n",
" leaky_re_lu_6 (LeakyReLU) (128, None, None, None, 0 \n",
" 128) \n",
" \n",
" batch_normalization_9 (Bat (128, None, None, None, 512 \n",
" chNormalization) 128) \n",
" \n",
" conv3d_7 (Conv3D) (128, None, None, None, 442496 \n",
" 128) \n",
" \n",
" leaky_re_lu_7 (LeakyReLU) (128, None, None, None, 0 \n",
" 128) \n",
" \n",
" batch_normalization_10 (Ba (128, None, None, None, 512 \n",
" tchNormalization) 128) \n",
" \n",
" max_pooling3d_3 (MaxPoolin (128, None, None, None, 0 \n",
" g3D) 128) \n",
" \n",
" batch_normalization_11 (Ba (128, None, None, None, 512 \n",
" tchNormalization) 128) \n",
" \n",
" global_average_pooling3d ( (128, 128) 0 \n",
" GlobalAveragePooling3D) \n",
" \n",
" dense (Dense) (128, 256) 33024 \n",
" \n",
" leaky_re_lu_8 (LeakyReLU) (128, 256) 0 \n",
" \n",
" batch_normalization_12 (Ba (128, 256) 1024 \n",
" tchNormalization) \n",
" \n",
" dropout (Dropout) (128, 256) 0 \n",
" \n",
" dense_1 (Dense) (128, 1) 257 \n",
" \n",
"=================================================================\n",
"Total params: 915921 (3.49 MB)\n",
"Trainable params: 913969 (3.49 MB)\n",
"Non-trainable params: 1952 (7.62 KB)\n",
"_________________________________________________________________\n"
]
}
]
},
{
"cell_type": "code",
"source": [
"import numpy as np\n",
"\n",
"num_samples = 1000\n",
"depth = 32\n",
"height = 64\n",
"width = 64\n",
"\n",
"X_train = np.random.rand(num_samples, depth, height, width, 1)\n",
"\n",
"y_train = np.random.randint(0, 2, size=(num_samples, 1))\n",
"\n",
"from sklearn.model_selection import train_test_split\n",
"X_train, X_val, y_train, y_val = train_test_split(X_train, y_train, test_size=0.2, random_state=42)\n"
],
"metadata": {
"id": "tJCxfMWEi9eV"
},
"execution_count": 6,
"outputs": []
},
{
"cell_type": "code",
"source": [
"history = model.fit(X_train, y_train, batch_size=32, epochs=2, validation_data=(X_val, y_val))"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "d7mP7B_DjBqM",
"outputId": "cbe8173c-2387-4d2f-a097-1cb4cfde0f25"
},
"execution_count": 7,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Epoch 1/2\n",
"25/25 [==============================] - 853s 34s/step - loss: 0.1433 - accuracy: 0.4888 - val_loss: 0.0215 - val_accuracy: 0.5050\n",
"Epoch 2/2\n",
"25/25 [==============================] - 832s 33s/step - loss: 0.0597 - accuracy: 0.5600 - val_loss: 0.0501 - val_accuracy: 0.5050\n"
]
}
]
},
{
"cell_type": "code",
"source": [],
"metadata": {
"id": "NXj_aacpjE_x"
},
"execution_count": null,
"outputs": []
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment