This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass
What I Want to Build
The idea is to build an app that runs locally in an individual's mobile. Suppose you go out and see a flower and have no idea what its name is, you take 2-3 pictures of the flower from various angles and send in the app.
The model gives you information about the flower like its name, its scientific name, when and where it is commonly found, etc.
Learning about a flower earns you a streak and and everyday if you upload your pic for different flowers it helps you build streak.
Why I Want to Build
Now, majority of us is either working or studying and we have lost that connection with nature. We spend most of the time indoors, so the purpose is to make individuals go out on walks or hiking trails on weekends thus helping with reinforcing a connection with nature and to know more about the most beautiful creation in the world, flowers.
How I Want To Build
I still need to learn on which open weight model will suit the best, learn how to fine-tune models, and I have a couple of features planned for this project.
Note
I wanted to participate but I was busy organising my own fest but I still posted the idea. Hopefully I work on it and come back here to share the code and demo.
Thanks for reading!
Top comments (2)
This is a nice use of a streak mechanic because the reward is tied to looking more closely at the world, not just opening the app. For an early version, I would make uncertainty part of the result: show a few likely species, explain what visual detail would disambiguate them, and allow “unknown.” Also make photo location an explicit choice, since exact locations can be sensitive for people and for rare plants.
Hi Alex,
Thank you for sharing your feedback, I am yet to work on it and it is an idea I have had for a long time now. Also, I would not reveal the exact location since what you said is absolutely true. I would keep the location limited to country and the states for rare species.