Splunk AR: Object Detection

The Splunk Augmented Reality (AR) team is excited to share more with you. In our first AR post, "Splunk AR: Taking Remote Collaboration To The Future is Already Here," from .conf20, we talked about our new Remote Collaboration feature, which helps field workers and remote experts collaborate in AR. In today’s post, we'll talk about our advancements in Object Detection. This new feature makes it even easier to deploy Splunk AR with your assets. 

Asset tags are the traditional method of recognizing assets with Splunk AR. This method supports importing your existing asset repository and deploying Splunk AR without any change management. However, we recognize that rolling out asset tags to those assets that aren’t already managed is expensive and time-consuming. We also know some assets may be tagged in physical locations that aren’t conducive to the AR user — good luck trying to squeeze in behind that vending machine! We have been working hard on Object Detection in Splunk AR to eliminate these issues. We now support recognizing assets in two ways: Text Detection and Logo Detection. Let’s watch this new technology in action!

It’s really easy to set up assets with Object Detection. You might be wondering how does it differentiate between multiple assets that have the same text label or logo? Location identifiers solve this unique challenge by disambiguating between visually identical assets. You have the choice between beacons and geofences.

Beacons are Bluetooth Low Energy devices that broadcast a unique identifier that Splunk AR recognizes when the phone is in close proximity. Once Splunk AR is in range of a beacon, it can successfully perform Logo and/or Text Detection on any nearby assets. Geofences are GPS-based boundaries indicating a “coverage area” for a given asset. Splunk AR uses the device’s location to determine which assets are nearby and then performs Logo and/or Text Detection. Geofences are a great choice for assets that are in distinct geographic locations since they don’t require any hardware to implement. However, when multiple assets share the same geographic location, beacons are the better choice because they offer more granular proximity detection.

Example logo for Logo Detection	Example text for Text Detection

Let’s take a look at an example of when to use each:

Geofences 

• Setting: College campus
• Use Case: Facilities team will scan room number to discover if the room needs cleaning.
• Splunk AR Scan Method: Text Detection on room number.
• Location Constraints: Room numbers are reused in different buildings (i.e. Room “100” is in several buildings).
• Location Identifier: Geofences. Each building maps to a geofence such that there are no duplicate room numbers inside any geofence.
   

Beacons 

• Setting: Factory floor
• Use Case: Maintenance person scans “Buttercup” branded logo on any machine to see service metrics.
• Splunk AR Scan Method: Logo Detection on Buttercup logo.
• Location Constraints: There are several machines in the same building all with the Buttercup logo. However, each machine is Bluetooth enabled and can act as a beacon.
• Location Identifier: Beacons. Splunk AR easily disambiguates between each beacon because beacon proximity is accurate within a 1 meter radius.
   

Setting up object detection in Splunk AR is fast and simple. Here are some next steps for you to try it out:

• Sign up for the AR Newsletter: Hear from us about new use cases and product updates.
• Splunk AR Documentation: Get started with Splunk AR.
• Reach out directly: Collaborate with us on your AR business problem immediately.
   

This article was co-authored by Sammy Lee, Product Manager for AR, Devin Bhushan, Engineering Manager for AR, and Jesse Chor, Head of Mobile Engineering. 

----------------------------------------------------
Thanks!
Jesse Chor