The difference between brand-based and non-marker augmented reality

Question

I am completely new to AR, and I have searched the Internet for marker and marker AR, but I am confused with marker and without marker AR. Suppose that an AR application runs an AR action when it scans for certain images. This is an AR or marker without AR. Isn't that an image? In addition, an AR marker is used to place the contents of the AR, using the accelerometer and device compass, as in referenceless AR?

Answer 1

In an application based on AR markers, images (or corresponding image descriptors) to be recognized are provided in advance . In this case, you know exactly what the application will look for when receiving camera data (camera frames). Most modern image recognition AR applications are brand-based. What for? Because it’s much easier to detect things that are hardcoded in your application.

On the other hand, an application without an AR token recognizes things that were not previously provided to the application in advance . This scenario is much more difficult to implement, because the recognition algorithm running in your AR application must identify patterns, colors, or some other functions that may exist in camera frames. For example, if your algorithm is capable of identifying dogs, this means that the AR application will be able to trigger AR actions when the dog is detected in the camera frame, without having to provide images to all dogs in the world (this is exaggerated, of course, preparing a database, for example) when developing applications.

In short: in an AR application on a marker basis, where image recognition is involved, the marker can be an image or corresponding descriptors (functions + key points). Typically, the AR marker is a black and white (square) image, such as a QR code. These markers are easily recognized and tracked. > To perform recognition (and possibly tracking), not much processing power is required on the end device.

There is no need for an accelerometer or compass in a marker-based application. The recognition library can calculate the placement matrix (rotation and translation) of the detected image relative to the camera of your device. If you know this, you know how far the image is recognized and how it rotates relative to your device’s camera. And from that moment AR begins ... :)

Answer 2

Well. Since I became without explanation. Below are a few details about mirrorless tracking:

In fact, there are several possibilities for augmented reality without “visual” markers, but none of them called targetless tracking.

The display of virtual information can be triggered by GPS, speech, or simply turning on the phone.

In addition, people tend to mix NFT (natural function tracking) with markerless tracking. With NFT, you can take the real picture as a marker. But it is still a marker.

This site has a good overview and a few examples for each marker: Types of markers This is mostly in German, but beware.

What you call mirrorless tracking today is a method that is best followed using Hololens (and its own programming language) or AR-Framework Kudan , Markerless Tracking does not find anything at its discretion. Instead, you can place an object at runtime somewhere in your field of vision. Marker tracking is then used to keep this object in place. It most likely uses a combination of sensor inputs and a solution to the SLAM problem (simultaneous localization and matching) at runtime.

EDIT: small update. It seems that hololonets create their own internal geometric representation of the room. 3D objects are then placed in this virtual room. After that, the room is stored in synchronization with the real world. The exact technology behind which it seems unknown, but some believe that it is based on Xbox Kinect technology.

Answer 3

Let it be simplified:

• Enhanced reality with a marker is when the tracked object is a black and white square marker. A good example that is really easy to see is shown below here : https://www.youtube.com/watch?v=PbEDkDGB-9w (you can try it yourself)
• Without a marker, augmented reality is when the tracked object can be anything: a picture, a human body, a head, eyes, a hand or fingers, etc., and in addition, you add virtual objects.

To summarize, location and orientation information is an important thing for augmented reality, which can be provided by various sensors and methods for them. If you have this information accurate, you can create some really good AR applications.

Answer 4

There seems to be some confusion between marker tracking and natural function tracking (NFT). Many AR SDKs support tracking as Markerless (NFT). This is still marker tracking as it uses a predefined image or feature set. This is not necessarily a black and white AR Toolkit marker type. For example, Vuforia uses NFT, which still requires a marker literally. In addition, in the most literal sense, tracking hands / face / body is also tracking markers, because the marker is a form. A marker inherent in a name does not require prior knowledge of the world or any form or object to track.

You can learn more about how markerless tracking was achieved here and see a few examples of markerless and markerless tracking here .

Answer 5

Based on the AR marker, a camera and a visual marker are used to determine the center, orientation and range of its spherical coordinate system. ARToolkit is the first full-featured marker tracking toolkit.

Markerless Tracking is one of the best ways to track right now. It performs active tracking and recognition of the real environment on any type of support without the use of special placed markers. Allows a more complex application of the concept of augmented reality.