I wanted to work with a Raspberry Pi (3) and do something with RPN (Reverse Polish Notation).
What it does
Captures an image of some arbitrary RPN equation/expression, sends to Google Vision API, detects any text, parses the result, and evaluations the resulting RPN expression(s).
How I built it
Capture the image with the Pi in Python, send to Vision API with Node.js, parse the results, and evaluation the resulting RPN expression(s).
Challenges I ran into
Overall consistency when capturing images and sending to Vision API was something erratic regarding the response. In general, high-contrast against a matte-background using computer-generated (i.e. typed) equations proved to be the most consistent.
Accomplishments that I'm proud of
That it all generally works as expected. Using Node.js and Python gave a lot of higher-level abstractions making it a much easier job to plug and play between the APIs so time could be focused on optimizing the environment to capture images that are well detected and handle parsing, validating, and evaluating the text.
What I learned
More experience using Python in a RPi environment and the basics of using Google Vision API.