How to improve the detection of small images

Question

I am doing a proof-of-concept with Sikuli and SeeShell for a larger RPA project. The problem where I get stuck is how to find small images inside the applications reliably. Some example images (icons mostly):

   

For a specific machine I can solve this by fine tuning the confidence level, but this is a very brittle solution that breaks easily (and it breaks for sure once we roll out the scripts to other machines).

Answer 1

Well there are a couple of things you can try

1. Use short-keys to reach to Icons if available.
2. Also, use tab or enter keys to deal with this

You can also use ClickRelative, which is a feature added by seeshell. This feature allows you to visually define an anchor area with a green box, and then mark the "click area" relative to it with a pink box. For small images, this makes the automation scripts much more error free.