This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.
Think about how often we scroll past pictures of cute birds, beautiful flowers, butterflies, and fascinating insects on our phones.
We stop to admire them. We like the picture. Maybe we share it with a friend. Then we keep scrolling.
But have you ever thought about how many of those things are living just a few steps away from you?
Imagine walking past the same garden every day, your eyes fixed on your phone. You might never notice that the garden is home to dozens of different flowers, birds visiting the trees, butterflies moving between plants, and tiny insects living in places you rarely look.
You don't need to travel to a forest or visit a national park to experience nature. Sometimes, an entire world of discoveries is waiting in a garden you pass every morning.
We have become so used to experiencing nature digitally that we sometimes forget to notice it in real life.
I wanted to build something that could change that.
What if, the next time you walked past a flower, instead of scrolling through another post, you stopped for a moment, took a picture, and discovered what it actually was?
What if a walk became a small adventure? What if you had a reason to explore a little more every day, notice a species you had never seen before, and learn something about the environment around you?
That's the idea behind Wildling — an AI-powered nature explorer designed to turn everyday walks into opportunities for discovery.
🌱 Meet Wildling
Wildling encourages you to step away from passive screen time and actively explore the nature around you.
The idea is simple: don't just scroll past nature. Go out and discover it.
Here's what you can do with Wildling.
📸 1. Discover and identify species with AI
You notice an unfamiliar flower, a bird sitting on a wall, or an insect resting on a leaf. Instead of wondering what it is and forgetting about it, open Wildling and take or upload a picture.
Wildling uses AI to compare the image with its species catalogue and suggest possible matches.
The current catalogue contains 921 species, covering categories such as plants, birds, insects, fungi, mammals, reptiles, amphibians, fish, and more.
You don't need to be a botanist, birdwatcher, or wildlife expert to start exploring. Curiosity is enough.
🔎 2. Explore possible matches
Sometimes identifying a species from a photograph isn't straightforward.
Wildling ranks candidate species and presents possible matches, giving you a starting point for learning about what you've found.
The goal isn't to pretend every identification is certain. The catalogue has limitations, and the model can make mistakes, particularly when a species isn't represented well.
Think of the result as an invitation to investigate further, not a scientific confirmation.
📚 3. Build your own nature collection
Every discovery can become a memory.
Wildling includes a collection feature that lets you keep track of the species you've explored and revisit your discoveries.
Over time, your collection can become a personal record of your encounters with the natural world.
Instead of remembering only the posts you've liked, you can look back at the things you've actually found around you.
🔥 4. Maintain your exploration streak
Want to make spending time in nature a regular habit?
Wildling includes streak tracking to encourage you to return to your exploration journey.
The idea is to build consistency through small actions. You don't need to spend hours outdoors every day. Even a short walk where you deliberately notice something new can be a meaningful start.
🏅 5. Earn badges as you explore
Wildling includes badges and progress features that make discovering nature more rewarding.
As you continue exploring, these features give you milestones to work toward and reasons to keep learning.
Instead of rewarding how long you remain on an app, the intention is to reward curiosity and exploration.
🧭 6. Take on quests
Wildling also includes quests that encourage you to approach nature exploration with a goal.
Rather than walking past the same surroundings without noticing them, you can use a quest as a prompt to look more carefully and seek out new discoveries.
It turns the question from “What should I do on my phone?” into “What can I discover outside today?”
🎨 7. Share your discoveries
Found something interesting? Wildling includes a sharing feature that lets you create shareable cards from your discoveries.
You can share what you've found with friends and inspire them to look more closely at the nature around them, too.
Imagine a friend sharing a picture of an unfamiliar flower they discovered in their neighbourhood. It might encourage you to investigate the plants growing along your own street.
That is the kind of interaction I want to encourage: sharing discoveries that bring our attention back to the real world.
🌍 The bigger picture
All these features work toward the same goal.
Wildling isn't trying to replace one endless feed with another. It's meant to give you a reason to open the app, explore something in the real world, learn from the experience, and then put your phone away again.
You might start by identifying one flower. Tomorrow, you might notice a bird. A few days later, you might find yourself looking more carefully at a garden you've passed hundreds of times.
The surroundings haven't changed. Your attention has.
Try Wildling
Live demo: https://huggingface.co/spaces/LasyapriyaPutti/wildling
Take a look at the app and try uploading a nature photograph to see what the model suggests.
💻 Code
GitHub repository: [https://github.com/lasyapriyaputti/wildling.git]
Wildling is built with React and Vite, with Python tools for preparing the species catalogue and embeddings.
🛠️ How I Built It
I wanted to make Wildling more than a static collection of species information. I wanted people to be able to photograph something they encountered and use AI to explore what it might be.
For the recognition system, I chose BioCLIP, an open-weight vision model designed for biological image understanding.
Here's how the main components work together.
1. Preparing the species catalogue
I prepared a catalogue of 921 species and generated text embeddings for the catalogue entries.
These embeddings act as reference representations. The application compares an image representation with the species embeddings to find candidate matches.
Keeping the catalogue and embeddings aligned is essential because each embedding must correspond to the correct species.
2. Bringing BioCLIP into the browser
The model's image encoder was exported to ONNX and integrated with ONNX Runtime Web.
This allowed the application to run image inference in the browser instead of requiring a dedicated remote inference API for each prediction.
3. Connecting images to species
When a user submits an image, Wildling preprocesses it and passes it through the image encoder.
The resulting image representation is compared with the catalogue embeddings. The application ranks the candidate matches and displays the suggested species.
The current system is catalogue-based, so its predictions depend on the species represented in the catalogue. A high-ranked match should not be treated as a calibrated probability or guaranteed identification.
4. Building the exploration experience
The frontend uses React and Vite. Around the recognition pipeline, I built the collection, streak, badge, quest, and sharing experiences to encourage people to return to nature exploration.
5. Testing and deployment
Getting an AI model working in a browser involved more than connecting a few components.
I had to validate the embeddings, check the ONNX model, manage the browser runtime files, and make sure the production build could load the model and generate predictions.
The image encoder is approximately 346 MB, which also made deployment and large-file management important parts of the work.
I deployed Wildling on Hugging Face Spaces and tested that the live application could run its browser-based recognition flow.
Wildling is still an early version. The catalogue needs continued validation, and the recognition system needs broader testing with real-world photographs. Those are important next steps for making the experience more reliable.
Why Does Open Innovation Matter?
Open innovation made Wildling possible in a way that would have been much harder if I had needed to build everything from scratch.
Building on existing research
BioCLIP gave me access to a model designed for biological image understanding. Instead of first training a vision model myself, I could focus on integrating the model into a useful application.
For an individual student developer, that access makes it possible to experiment with specialized AI using resources that would otherwise be difficult to obtain.
Running inference in the browser
Using ONNX Runtime Web allowed me to build a browser-based inference pipeline without relying on a separate hosted inference API for every image.
This gives the application more control over image processing and offers a path toward keeping images local. The model and required assets still need to be downloaded, so the whole experience is not automatically offline.
Making the project open to improvement
Open tools make it easier for developers to understand an implementation, test it, and experiment with different approaches.
Wildling can benefit from continued improvements to its catalogue, evaluation process, recognition pipeline, and user experience.
Open innovation doesn't mean the work is finished or every result is correct. It means more people can learn from the work and help make it better.
🌿 A Small Change in What We Notice
The next time you walk past a garden, you might be tempted to reach for your phone and start scrolling.
But what if you looked up instead?
What if you noticed a flower you've never seen before, followed a butterfly for a few seconds, or wondered which bird was making that sound?
You don't need to become a wildlife expert. You don't need to travel somewhere extraordinary. You just need to start noticing what's already around you.
I built Wildling because I believe AI can be useful in ways that don't demand more of our attention, but help us direct it toward things that matter.
The goal isn't to spend more time using Wildling. It's to spend more time discovering the world that exists beyond it.
So the next time you're walking past that familiar garden, put the scrolling on pause.
There might be a whole world waiting for you to notice it.
🌱 Wildling: Discover more. Scroll less. Touch grass.
#devchallenge #hf26challenge #opensource #ai #webdev
Top comments (0)