I’ve decided that before I go on I need to do a few things to this project. Well, technically I’m doing two tasks at once. Ok, ok. If you know me I’m only telling you about two tasks right now, but I’m really doing like 400.
The first task is that I’m just going to take the framework I have which was gleaned from previous projects and hacked together into something I like and strip it apart into an Angular (7 as of this writing) app. I’m not a fan of having another server running and putting up MongoDB isn’t going to make me happy on the processor front. I’m going to wire it to some database eventually, but for now, it’s going to be a flat-file JSON. I could leave an AngularJS version. However, I see a lot more Typescript code coming so I think it’s easier to use Angular now than integrate later on. Standalone tests will still work as they are.
The second task is that I’m going to keep converting items from my library over to javascript. As well, I keep importing opencv4nodejs examples as page examples. These both will help later on as I change the architecture over I’ll have the examples running and still be making progress.
In the video below you’ll see some examples of OpenCV not including the streaming video. I had issues with the video not realizing when I plugged in the USB camera to test that restarting Parallels would make that the default. These are not my examples, yet. I’ve discussed those examples in previous posts. My Tesseract examples aren’t processing correctly with either version of Tesseract I’m using for Node.js (I’ve discussed this in previous posts. The text box is so I can type commands to test them out instead of speaking so late at night I don’t wake everyone up. You’ll also notice that I say “Bob” instead of “Serinda” or some other name. That’s because “Serinda” is uncommon and I get varying results like “ok surrender.” For speed of testing, I’ve just gone with Bob (Bob Jones is my original Chuck Finley). One last note, I had to be very specific with “SIFT detection” and annunciate. When I incorporate other STT engines the accuracy will change. This is good enough.
Here is a short (and super shaky [sorry]) clip of what SERINDA does with voice and a couple items I still click because I’ve not wired them for voice yet.
One of these days soon, I’ll publish the code for this in the repository. It’ll all be OSS MIT license with attributions so no worries there. I just want to have something installable, stable, and further along, before people ask me a bunch of questions about all of that.
IN MY MIND BY WINK DOUBLEGUNS 