EcoPets — Recycle in the Real World. Raise a Hero in the Digital One.

Inspiration

EcoPets did not begin with a list of features. It began with a frustration.

Most people already know that recycling matters, yet awareness alone rarely creates lasting habits. Recycling can feel repetitive, the nearest suitable bin is not always easy to find, and the environmental value of a single action often feels distant or invisible.

There is usually no immediate feedback, no visible progression, and no compelling reason for someone to repeat the action the following day.

As a student deeply interested in electrical and computer engineering, programming, artificial intelligence, and environmental innovation, I had previously explored how intelligent technology could support more sustainable resource management. That experience showed me that technology can optimize environmental systems—but it also led me to a more difficult question:

How can technology motivate people, not merely inform them?

I began imagining a platform that would connect real environmental actions with the exploration, emotional attachment, and progression found in successful games.

If a location-based game could motivate millions of people to walk through their cities, discover new places, compete, and care about virtual creatures, perhaps the same mechanics could be redirected toward something with a measurable real-world purpose.

That became the foundation of EcoPets:

Every verified recycling action should create immediate, visible, and emotionally meaningful progress.

When I shared the idea with my friend and teammate, Ioanna Serf, we immediately recognized its potential. Together, we developed EcoPets as a team project combining environmental responsibility, artificial intelligence, geolocation, and digital pet evolution.

We did not want to build another application that simply tells people to recycle. We wanted to create a world that gives them a reason to do it repeatedly.

What EcoPets Does

EcoPets is a mobile-first, location-based sustainability game that transforms real-world recycling into digital progression.

The central EcoPets experience follows a simple but powerful loop:

Discover → Recycle → Scan → Verify → Earn → Evolve

Users can:

• Discover nearby recycling points through an interactive map.
• Identify bins according to the materials they accept.
• Capture or upload a photograph of a recyclable object.
• Verify the material through live AI-powered image recognition.
• Receive a clear classification result in real time.
• Earn progress after a successful recycling scan.
• Unlock digital creatures known as Eco-Pets.
• Build a collection that reflects their real environmental actions.

Each Eco-Pet is connected to a different recycling category:

• Glassy represents recycled glass.
• Papero represents paper and cardboard.
• Plasticko represents recyclable plastic.

When the AI successfully identifies a recyclable material, the corresponding Eco-Pet is unlocked.

Instead of offering abstract points with no emotional meaning, EcoPets creates a connection between the user and a digital companion. Each pet becomes a visible representation of the player’s environmental contribution.

The result is a new motivational model:

Recycle-to-Evolve

The user is not simply disposing of an object. They are completing a mission, receiving immediate feedback, expanding their collection, and contributing to a cleaner real-world environment.

How We Built It

We developed EcoPets as a responsive, mobile-first web application designed to work across desktop and mobile devices.

Frontend

The frontend was built using:

• React
• TypeScript
• Vite

We organized the application into reusable components and dedicated feature areas for:

• The recycling map
• The AI scanner
• Eco-Pet collection and unlocking
• Player progress
• XP and rewards
• User profiles
• Mobile-friendly navigation

This component-based structure allowed us to develop different areas of the application independently while maintaining a consistent interface and visual identity.

Backend

The backend was developed using Python and Flask.

It is responsible for:

• Receiving uploaded images
• Validating file types and image sizes
• Communicating securely with the AI service
• Processing the model’s response
• Returning structured verification data to the frontend
• Protecting sensitive credentials from browser exposure

The API key remains on the server and is never included in the public frontend code.

Live AI Verification

For material recognition, we integrated the Gemini Vision API.

The AI verification system is fully operational in the current version of EcoPets. It does not rely on simulated classifications or predetermined results.

When a user captures or uploads a real photograph, the image is securely sent to the Flask backend and analyzed live by Gemini.

The system returns structured information including:

• The detected material
• Whether the object appears recyclable
• A confidence score
• A short explanation of the result

The model classifies the image into one of four controlled categories:

• Paper
• Glass
• Plastic
• Unsupported or other material

The structured format allows the application to convert the AI response directly into game logic.

When a recognized recyclable material is verified, the corresponding Eco-Pet is unlocked. A glass object can unlock Glassy, paper can unlock Papero, and plastic can unlock Plasticko.

The complete core experience works from end to end:

1. The user captures or uploads a real image.
2. The React frontend sends the image securely to the Flask backend.
3. Gemini analyzes the image in real time.
4. The recyclable material is identified.
5. The result is validated and displayed to the user.
6. The matching Eco-Pet is unlocked.

Artificial intelligence is therefore not an additional feature placed on top of the project. It is the bridge connecting a physical object in the real world with progression inside the EcoPets experience.

Without verification, EcoPets would only be a game inspired by recycling. With live verification, the player’s real-world action becomes part of the gameplay itself.

Challenges We Faced

Making AI Verification Reliable

Real-world recycling photographs are unpredictable.

Lighting conditions, shadows, camera angles, damaged packaging, complex backgrounds, and objects made from several materials can all affect classification.

A plastic bottle may have a paper label. A cardboard package may contain plastic elements. A photograph may include several unrelated objects or provide an unclear view of the main material.

We had to design precise classification rules, constrain the model’s output, validate its response, and create understandable messages for unsupported or uncertain images.

This taught us that successfully integrating AI requires much more than sending an image to a model. The output must be structured, predictable, validated, and useful to the rest of the application.

Connecting the Complete Technical Flow

The scanner depends on communication between several different layers:

• The user’s device
• The React interface
• The Flask backend
• The Gemini API
• The Eco-Pet unlocking system

A small issue involving image formats, API requests, environment variables, response structures, dependencies, or cross-origin communication could interrupt the entire experience.

Testing each component independently was not enough. We had to test the complete journey—from uploading an image to receiving a verified result and unlocking the correct Eco-Pet.

Protecting the AI Integration

Because EcoPets communicates with an external AI service, sensitive credentials had to remain protected.

We routed all AI requests through the Flask backend instead of exposing the API key in the browser. We also added validation for uploaded files and designed controlled responses that the frontend could interpret safely.

This created a stronger and more secure foundation for the project.

Designing Fair Verification

A reward-based application must protect the integrity of its progression system.

The current version establishes the foundation through live AI material recognition. However, we also considered how future versions could prevent duplicate images, unrelated submissions, or repeated scans of the same object.

Possible future safeguards include:

• Duplicate-image detection
• Timestamp verification
• Geolocation confirmation
• Material-to-bin matching
• Verified recycling-point data
• Daily scanning limits for repeated objects

The challenge was to create an MVP that proves the core concept while designing the architecture with future anti-cheat mechanisms in mind.

Building a Focused MVP

EcoPets has the potential to include:

• Multiple Eco-Pet species
• Multi-stage evolutionary paths
• Daily missions and recycling streaks
• Friend systems
• Local and global leaderboards
• School competitions
• Community clean-up challenges
• Real-world sponsored rewards
• Municipal analytics
• Augmented reality
• Smart-bin integrations

Within the limited timeframe of a hackathon, attempting to implement every part of the vision would have weakened the central experience.

We therefore focused on proving the most important innovation:

A real recyclable object can be analyzed through live AI and transformed into meaningful digital progression.

Developing as a Team

Different parts of EcoPets were developed in parallel, including the interface, AI scanner, backend integration, game mechanics, and visual experience.

Bringing those elements together required clear communication, careful Git branch management, repeated testing, and a shared understanding of the project’s core purpose.

We learned that successful collaboration is not simply about dividing tasks. It is about ensuring that every technical decision contributes to the same product vision.

What We Learned

Building EcoPets taught us that a meaningful environmental mission is only the beginning.

For people to use an environmental product repeatedly, the experience must also be intuitive, trustworthy, rewarding, and enjoyable.

We learned that:

• AI should solve a central product problem rather than exist only as a technical showcase.
• AI responses must be structured and validated before they influence application state.
• Gamification becomes meaningful when rewards are connected to genuine actions.
• Emotional attachment can motivate users more effectively than abstract points.
• Users need immediate feedback to understand that their actions matter.
• Secure backend architecture is essential when working with external APIs.
• A focused and functional MVP is more valuable than many incomplete features.
• User feedback should influence product priorities rather than simply confirm the original idea.
• Successful teamwork requires communication, flexibility, and a shared vision.

Technically, we gained practical experience with:

• React architecture
• TypeScript
• Flask development
• API integration
• Live AI image analysis
• Image-upload workflows
• Structured model responses
• Responsive interface design
• Environment configuration
• Error handling
• Application state
• Git collaboration
• Connecting AI results to game mechanics

Most importantly, we learned how to transform an ambitious environmental concept into a working product that people can see, understand, and experience.

Accomplishments We Are Proud Of

We are especially proud that the central AI-powered workflow is completely functional.

A user can upload a real image, Gemini can analyze it live, the application can identify the recyclable material, and the correct Eco-Pet can be unlocked.

This is not simply a concept displayed in a presentation. It is a working connection between the physical and digital worlds.

We are also proud that EcoPets approaches sustainability through optimism rather than guilt.

Many environmental campaigns focus on fear, obligation, or the negative consequences of inaction. EcoPets takes a different approach by making environmental responsibility feel:

• Playful
• Personal
• Social
• Visible
• Rewarding

The user does not receive another reminder to recycle. They discover a world, complete a mission, care for a digital companion, and watch their positive actions become progress.

What’s Next

Our next steps include:

• Improving classification accuracy using more diverse recycling images.
• Adding duplicate-image and anti-fraud detection.
• Combining AI verification with geolocation and timestamps.
• Confirming that each material matches the selected recycling bin.
• Expanding the Eco-Pet collection.
• Introducing multiple evolution stages for every pet.
• Adding XP, daily missions, streaks, achievements, and badges.
• Creating friend systems and local leaderboards.
• Developing school and university competitions.
• Launching neighborhood and citywide recycling challenges.
• Partnering with sustainable businesses to offer real-world rewards.
• Developing anonymized recycling heatmaps for municipalities.
• Exploring augmented-reality Eco-Pet interactions.
• Supporting future smart-bin and IoT integrations.

In the long term, EcoPets could help municipalities understand recycling activity through aggregated and anonymized insights. Schools could use it to make environmental education interactive. Responsible businesses could sponsor measurable sustainability challenges and reward users for positive action.

Our ambition is for EcoPets to become more than a recycling game.

We want it to become an environmental engagement platform through which citizens, schools, businesses, and municipalities can work together—turning responsible behavior into part of everyday digital culture.

Pokémon GO used GPS to inspire millions of people to explore the world around them. EcoPets uses GPS, AI, and the emotional power of digital companions to inspire a generation to protect it.

Our Core Promise

Every item recycled becomes progress. Every action creates impact. Every Eco-Pet tells the story of a cleaner world.

Built With

• css
• html
• python
• typescript