This solution contains tutorial applications for the Anki Resources SDK.
There are four examples for now.
This shows how to open the assets, list the sounds, and to play them.
This shows how to open the assets, list the sprite sequence, and to display the frames on the screen. This example includes some code to colourize the otherwise grayscale PNGs (using a "ColorMatrix"). Note: set the colour to white to leave the sprite sequences with colours alone.
An example of how to play the composite images -- the image layers which can contain sprite sequences. Note: Not composite images reference sprite sequences. Most of the ones are in the "Weather" groupings.
Note: these composite images often expect to layer sprite sequences on top of the eyes, which are are rendered by the procedural face module (which isn't emulated here...
An example of how to classify items in the images.
This demo window shows:
-
a live camera feed (from the computers default camera)
-
The cropped feed, so that the result fits within the aspect ratio;
-
The cropped is overlayed with indicators where it things "people" are Note: The grid orientation may be wrong.
-
The labels of what it sees are listed to the right. These depend on which classifier/detector is used.
In general, it does not recognize most things, and is very sensitive to coloration and contrast. For instance, it recognizes my black spatula, but not my chrome-handled spatula, nor bamboo spatula. (Mobilenet has a lot of irrelevant things it recognizes, so I've no clue if it is used in the pet detector/tracker.)
Some tips:
-
The hand recognizer needs to see all of the fingers and thumb, so you will have to spread your fingers a bit, but keep them in the visual frame.
-
The hand recognizer needs the hand to be against a dark surface -- or perhaps just a contrasting surface.
-
Objects often should be close as possible to the camera, but still in the frame
An example of how to employ the text localization strings in the resources/assets.