spatialcomputing.info

Hello! AR enthusiast and self-taught developer here.

I'm currently looking for any advice regarding a solo project that I'd like to carry out in my free time, with a possible outlook in my professional life (pls let me get that promotion).

The idea is the following:

1. User scans a text (paper or digital) using his/her smartphone (IOS, Android) camera
2. OCR reads the text real time
3. If text matches some (50) keywords, some images (icons, emojis, 3D models, we'll see) are shown on screen

The mechanism is similar to Google Translate's "live" translation, even thought I will simply adjunct images to the text rather than superimposing as with the case of Google.

I'd like to hear your opinion on:

1) What backend would you recommend? There are a few GitHub examples with Python that are similar to what I'm willing to make and match my current level of expertise, but I'm open to any possibility (after all, it's an excuse to learn something new):

• Sudoku Solver / 1
• Sudoku Solver / 2 (better)
• OpenCV + PyScript
  

2) Where can I deploy this? It has to be a web app for being universal, so my main concern is the latency due to the processing. Does AWS provide for something specific? Is there any low-latency free alternative?