Qualcomm praises Moto Z2 Force and X4, explains their "Landmark Detection" camera feature


Qualcomm has come up with a rather interesting press release. The American chipset maker hailed Motorola's Moto Z2 Force and X4 models as being the first commercially available smartphones to use Qualcomm's Neural Processing Engine (NPE) for "Landmark Detection", one of their distinctive camera features. 

Okay, so what does Landmark Detection do? Basically, you can use the camera app of the Z2 Force and X4 to identify more than 1,200 landmarks from around the world in near real-time. After the phone recognizes the notable site, additional information such as informational widgets for Google Maps and Wikipedia also appear. Below, you can see a visual representation of the feature.


The deep learning model behind Landmark Detection uses Qualcomm's NPE. The speed at which objects are recognized comes from the fact that the NPE distributes the workload to the GPU's of the X4 and Z2 Force. 

In case you're wondering, this Neural Processing Engine is a software framework that provides developers with tools that accelerate neural network workloads (like Landmark Detection) on mobile devices. It's made to work with Snapdragon 600 and 800 series chipsets. Developers can also choose to distribute these workloads to the Kryo CPU and the Hexagon Digital Signal Processor (DSP). This approach differs from Apple's A11 chip and Huawei's Kirin 970, as they feature dedicated hardware for such tasks. 

Qualcomm says that Landmark Detection is only one example of what's possible for this AI tool. It will be interesting to see if other OEMs also start employing the Neural Processing Engine for their mobile app experiences.

Related phones

Moto Z2 Force Edition
  • Display 5.5" 1440 x 2560 pixels
  • Camera 12 MP / 5 MP front
  • Processor Qualcomm Snapdragon 835, Octa-core, 2450 MHz
  • Storage 64 GB + microSDXC
  • Battery 2730 mAh

FEATURED VIDEO

2 Comments

1. sgodsell

Posts: 7568; Member since: Mar 16, 2013

It's cool that SoCs are starting to embed AI or neural processing engines on board. But I kind of like the Pixels approach with a separate chip that uses a 1/10 the power that the main SoC uses. Plus the Pixels 2's SoC operates at 3 teraflops, which is leaps faster than Qualcomms neural processing engine. Even the A11 SoC with its neural processing engine operates at only 600 gigaflops.

2. Brewski

Posts: 725; Member since: Jun 05, 2012

If I pay all that money to fly to Paris I don't need my phone to tell me I'm taking a picture of the Eiffel Tower... I probably figured that out before I ever even got close to it.

Latest Stories

This copy is for your personal, non-commercial use only. You can order presentation-ready copies for distribution to your colleagues, clients or customers at https://www.parsintl.com/phonearena or use the Reprints & Permissions tool that appears at the bottom of each web page. Visit https://www.parsintl.com/ for samples and additional information.