Thats a very cool idea, Im thinking every time you smoke you have the potential of meeting a new friend. I can see how that can go totally wrong, like have creeps stalking girls on it.
1 Like
How does the Dark web Rating system work?
How would I verify sellers?
Ill be getting the data from the app?
You have a background in computer vision? Traditional or ai?
Making an app that makes decisions based on picture input and that actually works is a huge task.
Are you gainfully employed as a software engineer? Most computer engineers design/build circuits, thereās way more money in that field than in software engineering.
I am employed as a software engineer. I had no idea most computer engineers design/build circuits,. software kinda took over. How much are they making circuits engineer. I work with aws and api mostly.
Computer engineering is a subset of electrical engineering but is generally heavily seeded with computer science course work.
A computer engineer could pretty easily apply for software engineering jobs but theyād be selling themselves short and not using their main skill set. Computer engineers start over 6 figures with just a bachelors because not many people are in the field
1 Like
When I was looking for work, software engineering jobs were prevalent.
2 Likes
Thatās understandable and also I think computer engineering jobs are less spread out, and less likely to be remote/flexible.
I think an app that was similar to pokemonās pokedex would be cool. Gotta smoke em all
My brain really hurts programing software all the timeā¦ What do you do for a living?
Thats actually very funny. Sounds like a cool Vr idea
Iām a programmer, thatās why I was surprised youāre not a student and looking for more workā¦ 
Haha I figured, what are you currently working on rn?
Fellow programmer here. Iād have had my degree this past spring had I not dropped out last winter. Even though programming was my bag for many years during college doing contract work, I find myself much more at home in the lab than I ever did in front of the terminal. Glad to see some fellow nerds on here! Also, anybody got any use for eyetracking? Iāve got a pretty sweet codebase I could tap into.
1 Like
what are you building in terms of eyetracking?
Itās not specific like that ie mites vs thrips or pm vs mold.
Edit: I guess I use it more for identifying forage plants, I donāt use it to diagnose usually so it might do more that Iām giving it credit for.
Itās made for ALL plants and trees (but it does know cannabis plants) and it does a really good job at telling what kind of plant it is whether u take a picture of leaves, or a plant, or even the stem sometimes
1 Like
supporting a Java based web app as part of a fairly large devops team.
Iām curious how that works. Most likely itās a trained convolutional neural net (ai). But if that was the case , all of the plants it can identify it would have to have previously seen.
Iām contrast, if it was a more traditional computer vision task ā¦ I canāt even imagine how cumbersome that would be. Youād need a database of characteristics and or actual images to try to compare to. Itād probably work like crap and the CNN would be easier to build in software.
I think every time somebody snaps a pic and it identifies āthis is cannabisā it adds that pic to the database of cannabis pics and determines thru AI if Itās looking at a leaf, stem, plant, fruit, nut, etc
I could be wrong they might just have some internal definitions but thatās how I see it as an end user and not a programmer
Thatās be risky but itād be consider unsupervised training, which is coming of age.
More likely, a person labels every picture and then feeds it into the CNN for training. If the CNN is capable of distinguishing enough unique characteristics of the image of a cannabis plant, it can build a profile of a cannabis plant and break it into āfeaturesā. all of the feature mapping happens in what is considered a black box, humans donāt understand the inside of the CNN. Technically itās a boat load of chain rule calculations (relatively simple algebra/calculus) but the # of calculations is enormous. this demonstrates the beauty and mystery of a CNN, we put in the picture and say this is a cannabis plant and the CNN learns how to make that distinction somehow āon its ownā .
This is a vast oversimplification but you wouldnāt want the network to label pics itself unless it was dead nuts reliable, which Iām sure itās not. You can probably find difficult edge cases for it or plants not in its database that it may try to identify anyway, for example.
Yes but overall this app surprises me with its accuracy where other similar apps fall short