Woodpecker holes has been a challenge for Power & Utilities Industry for decades. This happens worldwide and leads to unexpected power outages and fines on companies. It requires companies to deploy resources to physically look for woodpecker holes to take any corrective and proactive action. With Amazon Rekognition Custom Labels, companies can use the power of machine learning to detect the woodpecker holes proactively with less operational overhead. Companies will only need a quality data set of images to build machine learning models using Amazon Rekognition Custom Labels.
In this lab, you will learn to how to label the images with holes/no holes and use that data set to train models with Amazon Rekognition Custom Labels with no machine learning expertise required. Amazon Rekognition Custom Labels will detect whether there are holes in utility poles or not. Utility companies can use this information to send out appropriate repair teams before an outage or fine can occur.
To run this lab, you need to complete prerequisite section from Lab 1.
On AWS Management Console, under services search for “Amazon SageMaker”, look for an instance with name “RekNotebookInstance-” prefix.
Under Actions, click on “Open Jupyter”
Under the “Files” tab, Open “New” dropdown and click on “Terminal”
In the terminal tab, type
git clone https://github.com/aws-samples/amazon-rekognition-workshops.git
Open the “ObjectDetection” folder created under “Files” tab
Click on the ground_truth_object_detection_turtorial.ipynb notebook.
Follow the instruction in the notebook to understand how you can identify well known people.
In order to execute each cell, click on the “Run” button at the top or press “Shift+Enter”.
Make sure you use “conda_python3” kernel and execute all cells in order.