What I Built
Isn't it amazing to discover bird species you've never seen around before? Or find flowers growing in your home town that you didn't know existed? Well, Nature Hunt solves exactly that.
Nature Hunt shows you a list of birds, plants, and rocks based on your selected location. It motivates you to go out and discover them.
Demo
Here is a link to the demo. Unfortunately, this demo doesn't have the Gemma 3 gemma3:1b-it-q4_K_M open-weight model integrated, so you can check out the GitHub repository instead.
Link:
https://nature-hunter.onrender.com/
Screenshots
Loading Images
Discoverable Images
AI Explanation
Code
1noahark
/
Nature-Hunt
Searches Nature
How I Built It
I used SerpApi to search for images based on the location users entered, and I used Gemma 3 (gemma3:1b-it-q4_K_M) to explain what they are. I chose this model because it was the lightest model I could find that was able to run on my computer.
I created an Express.js server and integrated SerpApi so that I can access the images through the API I created. I also used Ollama to run the Gemma model I'm using.
React ---> Ollama --> Gemma 3 (gemma3:1b-it-q4_K_M)
|
v
Express.js ---> SerpApi
Why Does Open Innovation Matter?
I used Ollama to run an AI model locally, which gave me more control over the model. I can do whatever I want, whenever I want, with the local model I downloaded without depending on a closed API. I can run and experiment with it as I like.
Prize Categories
Best Use of Render — Render was used to host the frontend and backend of the web app.
Best Use of Gemma — gemma3:1b-it-q4_K_M is the model explaining the search results.
Best Use of SerpApi — SerpApi fetches the data that the open-weight model explains and the images that are displayed from Google.
Top comments (1)
tr.ee/dev-to
Some comments have been hidden by the post's author - find out more