TensorFlow for Poets 1 hourFree Rate Lab GSP077 Overview TensorFlow is an open source library for numerical computation, specializing in machine learning applications. In this lab you will learn how to install and run TensorFlow on a single machine, then train a simple classifier to classify images of flowers. Introduction This lab uses transfer learning to train your machine. In transfer learning, when you build a new model to classify your original dataset, you reuse the feature extraction part and re-train the classification part with your dataset. This method uses less computational resources and training time. Deep learning from scratch can take days, but transfer learning can be done in short order. The model for this lab is trained on the ImageNet Large Visual Recognition Challenge dataset. These models can differentiate between 1,000 different classes, like "Zebra", "Dalmatian", and "Dishwasher". In a production environment, you'll have a choice of model architectures, so you can determine the right tradeoff between speed, size, and accuracy for your problem. At the end of the lab you'll have the opportunity to retrain the same model to tell apart a small number of classes based on your own examples. What you will learn: How to use Python and TensorFlow to train an image classifier How to classify images with your trained classifier What you need: A basic understanding of Linux commands Your own images of flowers to upload into the lab Set up Before you click the Start Lab button Read these instructions. Labs are timed and you cannot pause them. The timer, which starts when you click Start Lab, shows how long Cloud resources will be made available to you. This Qwiklabs hands-on lab lets you do the lab activities yourself in a real cloud environment, not in a simulation or demo environment. It does so by giving you new, temporary credentials that you use to sign in and access the Google Cloud Platform for the duration of the lab. What you need To complete this lab, you need: Access to a standard internet browser (Chrome browser recommended). Time to complete the lab. Note: If you already have your own personal GCP account or project, do not use it for this lab. Note: If you are using a Pixelbook please open an Incognito window to run this lab. How to start your lab and sign in to the Console 1. Click the Start Lab button. If you need to pay for the lab, a pop-up opens for you to select your payment method. On the left is a panel populated with the temporary credentials that you must use for this lab. 2. Copy the username, and then click Open Google Console. The lab spins up resources, and then opens another tab that shows the Choose an account page. Tip: Open the tabs in separate windows, side-by-side. 3. On the Choose an account page, click Use Another Account. 4. The Sign in page opens. Paste the username that you copied from the Connection Details panel. Then copy and paste the password. Important: You must use the credentials from the Connection Details panel. Do not use your Qwiklabs credentials. If you have your own GCP account, do not use it for this lab (avoids incurring charges). 5. Click through the subsequent pages: Accept the terms and conditions. Do not add recovery options or two-factor authentication (because this is a temporary account). Do not sign up for free trials. After a few moments, the GCP console opens in this tab. Note: You can view the menu with a list of GCP Products and Services by clicking the Navigation menu at the top-left, next to “Google Cloud Platform”. Create a development machine in Compute Engine In the Google Cloud Console, go to Compute Engine > VM Instances > Create. Configure the instance this way: Name: devhost Region: us-central1 Zone: us-central1-a Machine Type: 4 vCPUs (n1-highcpu-4) instance Boot disk: Click Change, then select Ubuntu 16.04 LTS, then click Select Identity and API Access: "Allow full access to all Cloud APIs" Click Create. This will serve as your development bastion host. SSH into devhost by clicking the SSH button on the Console. Now you're ready to install TensorFlow onto this machine. Test Completed Task Click Check my progress to verify your performed task. If you have completed the task successfully you will be granted with an assessment score. Create a development machine in Compute Engine Check my progress Install TensorFlow In the SSH window, run the following to install Python package manager: sudo apt-get update sudo apt-get install -y python-pip python-dev python3-pip python3-dev virtualenv Install and configure Virtual Environment for using Python 3.5 pip install virtualenv virtualenv -p python3 venv source venv/bin/activate Now install TensorFlow with the following: pip install --upgrade https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.14.0-cp35cp35m-linux_x86_64.whl Start Python in the SSH window: python At the Python prompt ( >>> ), import TensorFlow: import tensorflow as tf Note: Ignore depreciation warnings. Create a constant that contains a string: hello = tf.constant('Hello, TensorFlow!') Then create a TensorFlow session: sess = tf.Session() You can ignore the warnings that the TensorFlow library wasn't compiled to use certain instructions. Display the value of hello: print(sess.run(hello)) If successful, the system outputs: Hello, TensorFlow! Stop the Python interactive shell: exit() Test Completed Task Click Check my progress to verify your performed task. If you have completed the task successfully you will granted with an assessment score. Install TensorFlow Check my progress Clone the git repository All the code used in this lab is in this git repository. In the SSH window, run the following command to clone the repository: git clone https://github.com/googlecodelabs/tensorflow-for-poets-2 Now cd into it: cd tensorflow-for-poets-2 Test Completed Task Click Check my progress to verify your performed task. If you have completed the task successfully you will granted with an assessment score. Clone the git repository Check my progress Download the training images Before you start any training, you'll need a set of images to teach the model about the new classes you want to recognize. We've created an archive of creative-commons licensed flower photos to use initially. Download the photos (218 MB) by invoking the following commands in the SSH: curl http://download.tensorflow.org/example_images/flower_photos.tgz \ | tar xz You should now have a copy of the flower photos in your working directory. Confirm the contents of your working directory by issuing the following command: ls flower_photos Example Output: You should see the following objects: LICENSE.txt daisy dandelion roses sunflowers Move the flower images directory into tf_files: mv flower_photos tf_files Test Completed Task tulip Click Check my progress to verify your performed task. If you have completed the task successfully you will granted with an assessment score. Download the training images and move them to the tf_files directory Check my progress (Re)training the network Configure your MobileNet The retrain script can retrain either Inception V3 model or a MobileNet. In this lab you will use a MobileNet. The principal difference is that Inception V3 is optimized for accuracy, while the MobileNets are optimized to be small and efficient, at the cost of some accuracy. Inception V3 has a first-choice accuracy of 78% on ImageNet, but the model is 85MB, and requires many times more processing than even the largest MobileNet configuration, which achieves 70.5% accuracy, with just a 19MB download. The following configuration options are available: Input image resolution: 128,160,192, or 224px. Unsurprisingly, feeding in a higher resolution image takes more processing time, but results in better classification accuracy. We recommend 224 as an initial setting. The relative size of the model as a fraction of the largest MobileNet: 1.0, 0.75, 0.50, or 0.25. We recommend 0.5 as an initial setting. The smaller models run significantly faster, at a cost of accuracy. With the recommended settings, it typically takes only a couple of minutes to retrain on a laptop. For this lab you'll use the highest settings. In the SSH window, pass the settings inside Linux shell variables with the following commands: export IMAGE_SIZE=224 export ARCHITECTURE="mobilenet_0.50_${IMAGE_SIZE}" The graph below shows the first-choice-accuracies of these configurations (yaxis), vs the number of calculations required (x-axis), and the size of the model (circle area). 16 points are shown for MobileNet. For each of the 4 model sizes (circle area in the figure) there is one point for each image resolution setting. The 128px image size models are represented by the lower-left point in each set, while the 224px models are in the upper right. An extended version of this figure is available in slides 84-89 here. Open a firewall for Tensorboard on port 6006 TensorBoard is a monitoring and inspection tool included with TensorFlow. You will use it to monitor the training progress. In order for your VM to get to TensorBoard, you'll create a firewall rule. In the Console, click on devhost. Scroll down and under Network Interfaces, click on default under the Network heading. Click on Firewall Rules on the left, then "Create Firewall Rule" at the top of the page. Now set up your firewall: Name: tensorboard Targets: All instances in the network Source IP ranges: 0.0.0.0/0 Protocol and ports: check tcp, and then type "6006" Test Completed Task Click Check my progress to verify your performed task. If you have completed the task successfully you will granted with an assessment score. Open a firewall for Tensorboard on port 6006 Check my progress Start TensorBoard In the SSH window, launch tensorboard in the background. tensorboard --logdir tf_files/training_summaries & Note: This command will fail with the following error if you already have a TensorBoard process running:ERROR:tensorflow:TensorBoard attempted to bind to port 6006, but it was already in use You can kill all existing TensorBoard instances with: pkill -f "tensorboard" This URL will be how you get to TensorBoard: http://devhost:6006 Replace devhost with the external IP address for your VM and refresh the browser. Now you're ready. Note: You can find the external IP for devhost in the VM instances screen (Navigation menu > Compute Engine > VM instances). Investigate the retraining script The retrain script is part of the tensorflow repo, but it is not installed as part of the pip package. For simplicity, it has been included it in the lab repository. You can run the script using the python command. Take a minute to skim its "help". You might need the press Enter again to return to the command line in the SSH window. python -m scripts.retrain -h Run the training As noted in the introduction, Imagenet models are networks with millions of parameters that can differentiate a large number of classes. We're only training the final layer of that network, so training will end in a reasonable amount of time. In the SSH, start your retraining with one big command (note the -summaries_dir option, sending training progress reports to the directory that tensorboard is monitoring): python -m scripts.retrain \ --bottleneck_dir=tf_files/bottlenecks \ --how_many_training_steps=500 \ --model_dir=tf_files/models/ \ --summaries_dir=tf_files/training_summaries/"${ARCHITECTURE}" \ --output_graph=tf_files/retrained_graph.pb \ --output_labels=tf_files/retrained_labels.txt \ --architecture="${ARCHITECTURE}" \ --image_dir=tf_files/flower_photos This script downloads the pre-trained model, adds a new final layer, and trains that layer on the flower photos you've downloaded to the lab. ImageNet does not include any of these flower species you're training on. The kinds of information that make it possible for ImageNet to differentiate among 1,000 classes are useful for distinguishing other objects. By using this pre-trained network, you're using that information as input to the final classification layer that distinguishes your flower classes. This will take several minutes to complete. Read through the next couple of sections while the training is running. More about Bottlenecks This section and the next provide background on how this retraining process works. The first phase analyzes all the images on disk and calculates the bottleneck values for each of them. What's a bottleneck? These ImageNet models are made up of many layers stacked on top of each other, a simplified picture of Inception V3 from TensorBoard, is shown above (all the details are available in this paper, with a complete picture on page 6). These layers are pre-trained and are already very valuable at finding and summarizing information that will help classify most images. For this lab, you are training only the last layer (final_training_ops in the figure below). While all the previous layers retain their already-trained state. In the above figure, the node labeled "softmax", on the left side, is the output layer of the original model. All the nodes to the right of the "softmax" were added by the retraining script. A bottleneck is an informal term often used for the layer just before the final output layer that actually does the classification. "Bottelneck" is not used to imply that the layer is slowing down the network, but because near the output, the representation is much more compact than in the main body of the network. Every image is reused multiple times during training. Calculating the layers behind the bottleneck for each image takes a significant amount of time. Since these lower layers of the network are not being modified their outputs can be cached and reused. So the script is running the constant part of the network, everything below the node labeled Bottlene... above, and caching the results. The command you ran saves these files to the bottlenecks/ directory. If you rerun the script, they'll be reused, so you don't have to wait for this part again. More about Training Once the script finishes generating all the bottleneck files, the actual training of the final layer of the network begins. The training operates efficiently by feeding the cached value for each image into the Bottleneck layer. The true label for each image is also fed into the node labeled GroundTruth. Just these two inputs are enough to calculate the classification probabilities, training updates, and the various performance metrics. As it trains you'll see a series of step outputs, each one showing training accuracy, validation accuracy, and the cross entropy: The training accuracy shows the percentage of the images used in the current training batch that were labeled with the correct class. Validation accuracy is the precision (percentage of correctly-labelled images) on a randomly-selected group of images from a different set. Cross entropy is a loss function that gives a glimpse into how well the learning process is progressing (lower numbers are better here). The figures below show an example of the progress of the model's accuracy and cross entropy as it trains. When your model has finished generating the bottleneck files you can check your model's progress by opening TensorBoard, and clicking on the figure's name to show them. TensorBoard may print out warnings to your command line. These can safely be ignored. A true measure of the performance of the network is to measure its performance on a data set that is not in the training data. This performance is measured using the validation accuracy. If the training accuracy is high but the validation accuracy remains low, that means the network is overfitting, and the network is memorizing particular features in the training images that don't help it classify images more generally. The training's objective is to make the cross entropy as small as possible, so you can tell if the learning is working by keeping an eye on whether the loss keeps trending downwards, ignoring the short-term noise. By default, this script runs 4,000 training steps. Each step chooses 10 images at random from the training set, finds their bottlenecks from the cache, and feeds them into the final layer to get predictions. Those predictions are then compared against the actual labels to update the final layer's weights through a backpropagation process. As the process continues, you should see the reported accuracy improve. After all the training steps are complete, the script runs a final test accuracy evaluation on a set of images that are kept separate from the training and validation pictures. This test evaluation provides the best estimate of how the trained model will perform on the classification task. You should see an accuracy value of between 85% and 99%, though the exact value will vary from run to run since there's randomness in the training process. (If you are only training on two classes, you should expect higher accuracy.) This number value indicates the percentage of the images in the test set that are given the correct label after the model is fully trained. Go back to TensorBoard tab When the command line returns that means that the training is complete. Go look at the TensorBoard tab again to see what your results look like. Using the Retrained Model The retraining script writes data to the following two files: tf_files/retrained_graph.pb, which contains a version of the selected network with a final layer retrained on your categories. tf_files/retrained_labels.txt, which is a text file containing labels. Classifying an image Next you will run the script on this image of a daisy: flower_photos/daisy/21652746_cc379e0eea_m.jpg Image CC-BY by Retinafunk python -m scripts.label_image \ --graph=tf_files/retrained_graph.pb \ --image=tf_files/flower_photos/daisy/21652746_cc379e0eea_m.jpg Each execution will print a list of flower labels, in most cases with the correct flower on top (though each retrained model may be slightly different). You should get results like this for the daisy photo: Daisy 0.99508375 Dandelion 0.0028086917 sunflowers 0.002093148 Roses 1.37025945e-05 Tulips 6.3252025e-07 This indicates a high confidence (~99%) that the image is a daisy, and low confidence for any other label. Try it again with a new image: flower_photos/roses/2414954629_3708a1a04d.jpg Image CC-BY by Lori Branham python -m scripts.label_image \ --graph=tf_files/retrained_graph.pb \ --image=tf_files/flower_photos/roses/2414954629_3708a1a04d.jpg How did this image do compared to the first one? (Optional) Trying other hyperparameters if you have time You can use label_image.py to classify any image file you choose, either from your downloaded collection, or new ones. You just have to change the -image file name argument to the script. The retraining script has several other command line options you can use. You can read about these options in the help for the retrain script: python -m scripts.retrain -h Try adjusting some of these options to see if you can increase the final validation accuracy. For example, the --learning_rate parameter controls the magnitude of the updates to the final layer during training. So far we have left it out, so the program has used the default learning_rate value of 0.01. If you specify a small learning_rate, like 0.005, the training will take longer, but the overall precision might increase. Higher values of learning_rate, like 1.0, could train faster, but typically reduces precision, or even makes training unstable.You need to experiment carefully to see what works for your case. It is very helpful for this type of work if you give each experiment a unique name, so they show up as separate entries in TensorBoard. It's the --summaries_dir option that controls the name in tensorboard. Earlier --summaries_dir=training_summaries/basic was used. TensorBoard is monitoring the contents of the training_summaries directory, so setting --summarys_dir to training_summaries or any subdirectory of training_summaries will work. You may want to set the following two options together, so your results are clearly labeled: --learning_rate=0.5 --summaries_dir=training_summaries/LR_0.5 (Optional) Training on your own categories After you see the script working on the flower example images, you can start looking at teaching the network to recognize different categories. In theory, all you need to do is run the tool, specifying a particular set of subfolders. Each sub-folder is named after one of your categories and contains only images from that category. If you complete this step and pass the root folder of the subdirectories as the argument for the --image_dir parameter, the script should train the images that you've provided, just like it did for the flowers. The classification script uses the folder names as label names, and the images inside each folder should be pictures that correspond to that label, as you can see in the flower archive: Collect as many pictures of each label as you can and try it out! Test your knowledge Test your knowledge on the Google Cloud Platform by taking our quiz. In transfer learning, when you build a new model to classify your original dataset, you reuse the feature extraction part and re-train the classification part with your dataset. This method uses less computational resources and training time. Deep learning from scratch can take days, but transfer learning can be done in short order. True False Congratulations! You've taken your first steps into a larger world of deep learning! Finish your Quest This self-paced lab is part of the Qwiklabs Google Cloud Solutions ll: Data and Machine Learning and Intermediate ML: TensorFlow on GCP Quests. A Quest is a series of related labs that form a learning path. Completing a Quest earns you the badge above, to recognize your achievement. You can make your badge (or badges) public and link to them in your online resume or social media account. Enroll in an above Quest and get immediate completion credit if you've taken this lab. See other available Qwiklabs Quests. Take your next lab Continue your Quest with Creating an Obect Detection Application using TensorFlow, or try one of these suggestions: Creating Custom Interactive Dashbosards with Bokeh and BigQuery Running Distributed TensorFlow on Compute Engine Next steps / learn more You can see more about using TensorFlow at the TensorFlow website or the TensorFlow Github project. There are lots of other resources available for TensorFlow, including a discussion group and whitepaper. If you make a trained model that you want to run in production, you should also check out TensorFlow Serving, an open source project that makes it easier to manage TensorFlow projects. If you're interested in running TensorFlow on mobile devices try the second part of this tutorial which will show you how to optimize your model to run on Android. Go have some fun in the TensorFlow Playground! Sign up for the full Coursera Course on Machine Learning Google Cloud Training & Certification ...helps you make the most of Google Cloud technologies. Our classes include technical skills and best practices to help you get up to speed quickly and continue your learning journey. We offer fundamental to advanced level training, with on-demand, live, and virtual options to suit your busy schedule. Certifications help you validate and prove your skill and expertise in Google Cloud technologies. Manual last updated July 29, 2019 Lab last tested June 6, 2019 Copyright 2020 Google LLC All rights reserved. Google and the Google logo are trademarks of Google LLC. All other company and product names may be trademarks of the respective companies with which they are associated.