Now This Is Deep

Depth cameras, which can measure the distance between objects in a scene, have diverse applications in a wide range of fields. They are often used in robotics to enable precise navigation and obstacle avoidance, in industrial inspection to allow for the precise measurement and analysis of products and machinery, in transportation to enable driver assistance systems and autonomous vehicles, and even in medical imaging for precision measurement and visualization of the human body. And this list is just scratching the surface of what can be done with a depth camera, so it is no surprise that these devices have been skyrocketing in popularity in recent years.

Unfortunately, a lot of the potential that could be achieved using this technology is being hindered due to the cost and complexities associated with integrating depth cameras into larger hardware designs. The high priced components that go into manufacturing many of these cameras leave them out of the price range of many consumer devices. Furthermore, purchasing a depth camera is only the beginning — from there, they may need to be set up and calibrated via complex procedures, after which the data that they produce needs to be analyzed by complicated computer vision algorithms on expensive hardware.

If just the thought of using a depth camera is enough to make your head spin, you are not alone. It may get a lot easier in the future, however, thanks to Orbbec's Femto Mega depth camera, which just debuted at this year's Consumer Electronics Show (CES) in Las Vegas. Orbbec partnered with Microsoft to develop the industry's highest resolution 3D camera with an integrated machine learning processing and analysis environment.

The camera is able to provide excellent quality in a low-cost device by leveraging Microsoft's ToF technology, which pairs a 4K resolution RGB camera with a 1 megapixel depth sensor. This very capable vision system feeds into the inexpensive, yet very powerful, NVIDIA Jetson Nano single board computer with a 128-core Maxwell GPU to accelerate processing of the captured image and depth data. The Jetson comes preloaded with the image processing algorithms to make integration with external systems as easy as possible.

Getting started with depth sensing can be as simple as plugging in a single ethernet cable that provides access to the device's data as well as power for the camera and processing unit. A set of APIs have been developed to further facilitate integration of the camera with custom device designs.

Orbbec's Femto Mega was designed to extend the technologies employed in HoloLens and the Azure Kinect Devkit for a wide range of industrial applications. The CTO of Speedcargo Technologies, a large-scale cargo digitization solution provider, explained that they are expanding their present Azure Kinect-based operations using the Femto Mega. It is enabling them to maintain compatibility with their current system, but is also helping them to cut costs, and to reduce system size and complexity. He points out that this will be pivotal in helping them expand their operations to new locations.

Orbbec will be demonstrating the Femto Mega, as well as their other 3D vision solutions, at their booth at CES. Be sure to swing by if you are in town for the show.