The Journey of an Aerial Image: How SkyX turns aerial images into actionable insights

Our VP of Marketing, Julie Ford, interviewed Jamie Alexander, our Head of Product, on how SkyVision processes and converts the images collected during pipeline monitoring missions into meaningful insights for our customers.

JULIE: What are the steps that an image goes through after being uploaded to SkyVision?

JAMIE: There are a number of steps that an image goes through in our image processing pipeline.

First, the flight crew captures the images with drones and then uploads them to SkyVision where they’re stored in a secure cloud.

Then, we take each image and validate it to make sure that it’s within the same proximity of the oil field or asset that it’s been taken from.

Next, we begin a series of validation steps. We make sure that there is no duplicate image already in the system. There are a lot of images coming in so we want to make sure an image hasn’t been uploaded multiple times in error. The duplicate checks involve checking the date the image was captured and the location data from the image.

Then, we pull out the metadata such as the pitch, roll and yaw of the drone, the focal length and resolution of the camera and using terrain models of the earth run a series of complex calculations to determine the actual geolocation of that image on the earth. We do this because the camera lens can never be perfectly parallel to the surface of the earth.   

Once we’ve run this process on all of the images , we collect and create what we call a “flight.” Then we create a preview image, which takes all the different points from the locations of the images and then stitches them together to form  a flight path.

From there, the flight is ready for analysis.

SkyX Aerial Image Processing Journey Infographic

JULIE: Why do the images need to go through this process? What’s the end goal?

JAMIE: First of all, images need to be stored in the system so that they can be reviewed. Second, we need to do the geolocation step to be able to locate the images on the earth, including latitude, longitude and center of the image. Then thirdly, we’re hired by our customers to monitor assets from above and the aerial imagery is what we use to help them visualize their assets.

JULIE: How many images can the system process at once?

JAMIE: We’ve built the infrastructure in such a way that it scales almost infinitely. We have the ability to scale our system to handle any volume of images. And these images are really big. They’re about 10 megabytes each.

We’re using Amazon Web Services because it’s got massive capacity. For the image processing, or long-running kinds of processes where it takes time to process an image and go through each step, we’re using something called AWS Lambda. Basically, Lambda scales automatically on demand. So, in theory, you can throw in 1,000 images at once and it can process them all at the same time and come back with the results.

JULIE: How long does it take to process an image from start to finish?

JAMIE: The processing time is really quick. It’s the upload time that can vary depending on the connection and bandwidth that the flight crew have when they’re uploading the images. When they’re in rural areas doing their flights it can take them some time to upload images. We built the system with safeguards and data recovery so that if the upload fails halfway through, it can pick up where it left off.

JULIE: How does this workflow benefit the customer?

JAMIE: Ultimately, what the customer wants is insights from us in the shortest time possible. Right now, we have an SLA with our customers to provide them with a report on urgent issues within 12 hours of inspection and all of the areas of interest identified along with their asset within five days of each flight. Speed is important. We want to be able to turn around the data and provide insights to the customer quickly.

On top of speed, the other big thing is to be able to do it in a systematic way. Many of our competitors and other service providers are doing one-offs. They’ll fly over an area, they’ll get the imagery, they’ll go through it manually, they’ll provide a report back to the customer, then they’ll have some discussion. It’s a one-and-done type of activity.

But, if you want to do it periodically – every quarter, or every month, every week, or even every day – and do it over and over again to be able to see how things have changed over time and then scale it up to multiple assets in different geographies, you really do need a system like SkyVision that helps bring it all together.

Learn more