ReWaste - AI Waste Sorter Indonesia

• 

  Snap a photo of waste -> know where it goes. AI waste sorter for Indonesia's 4-color SNI system.

• 

  Tailwind + JS + TF.js. Waste AI. Mock now, real next.

• 

  AI waste scan. 1,250 pts. Photo -> PET bottle -> Bin.

• 

  AI scans waste, 4 SNI bins sort, Semarang saves.

• 

  AI scan 92% PET bottle -> Yellow Bin. +25 pts. Reuse idea: Bottle Planter.

• 

  93.7% accuracy. AI sorts trash to 4 color bins.

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for ReWaste - AI Waste Sorter Indonesia

Inspiration

70% sampah di Indonesia tidak terpilah di sumber. Di Semarang, bank sampah menolak sampah campur, TPA Jatibarang overload. Kami ingin bikin solusi yang dipakai ibu rumah tangga dalam 5 detik: foto sampah -> tau tempat buangnya.

What it does

ReWaste - SampahPintar adalah AI waste sorter berbasis web:

• Upload foto sampah -> klasifikasi 4 kategori SNI Indonesia: Recyclable (Kuning), Organic (Hijau), B3/Hazardous (Merah), Residual (Biru)
• Menampilkan confidence score, cara cuci & pilah, dan warna bin yang benar
• Memberi ide reuse & rute donasi (sisa makanan -> Food Bank, kardus -> bank sampah)
• Impact tracker: kg sampah terdiversi, CO2 saved, poin komunitas

Built for Arbiter Hacks V1 Theme Track: Rethinking Waste - primary purpose adalah waste reduction, sorting, recycling.

How we built it

• Frontend: React-style UI + Tailwind CSS (prototype yang ada di demo)
• AI Vision (roadmap): TensorFlow.js + MobileNetV2 fine-tuned dengan dataset TrashNet + TACO + 500 foto sampah lokal Indonesia
• Mock logic sekarang untuk demo, next: real model on-device
• Storage: LocalStorage untuk impact tracking offline-first

Challenges we ran into

Dataset sampah Indonesia beda dengan dataset US/Eropa - plastik kresek kotor, sterefoam bekas makanan. Harus buat label yang sesuai SNI dan sistem warna Indonesia, bukan US single-stream.

Accomplishments that we're proud of

Working end-to-end prototype yang bisa dipakai orang awam, bukan cuma teknisi. Design mengikuti warna bin Indonesia asli, bukan template.

What we learned

Pentingnya contextual AI - model harus paham konteks lokal (bahasa, warna bin, infrastruktur bank sampah).

What's next for ReWaste

• Deploy real MobileNet model
• Integrasi peta bank sampah Semarang via OpenStreetMap
• Kolaborasi dengan DLH Kota Semarang dan komunitas
• Mode Bahasa Jawa & Indonesia

Track

Theme Track: Rethinking Waste

Built With

• css
• flow.js
• html5
• javascript
• netlify
• tailwild
• tensor