We went through a couple iterations of these images as shown below. The biggest issue was that each plot had to be distinctive in order for the augmented reality to register it (although that didn't matter in the end since we decided not to use augmented reality). 

Originals

The images started off with an icon style, with a circle framing the typical shape of the plant in the forms of seed , young plant, and fully grown plant. It was intended for a brown overlay to indicate a dead plant and a water drop background to indicate the need for water. However the plants for these designs were not the official choices and many aspects of the images were too similar to each other.

Trial 2

The official plant choices were made in Adobe Illustrator and made with circular frames that had different colors to define state. The shapes of these frames were hard for the camera to to define when rotated, so the designs had to be scrapped.

Trial 3

Aliya chose to use different shapes for each plant. By using ovals for one type, she hoped that the rotation would not have as much trouble with the system. However after testing them in the vuforia system, she found that the images were not good trigger designs and had to figure out a quick solution so that tests could be run with the programming.

Final Set

After reading through an advisory page for trigger image design on vuforia, Aliya found that the vuforia system did not look at the different colors of a trigger image, but the points where two lines met. The more individual points, the more identifiable the image. She placed the trial 3 images over different dirt backgrounds. Planting was given a moist soil background, Watering was given a rocky soil background, and Dead was given a cracked earth background. These triggers worked out well for the system.