Logo Lanfrica

AbrhamTesf/Adwa

Domaine:

natural language processing

Type de record:

softwareproject
Créateur:
Abr
Hôte:
An open-source digital heritage & WebAR tourism platform for the Adwa Victory Memorial Park. Preserving Ethiopian history through AI visual scan, interactive 3D virtual tours, and multilingual persona guides. # Adwa AI Companion & Heritage Platform > **Adwa is not only a place to visit; it is a story we are responsible for carrying forward.** Ethiopian and global open-source developers: help make the history of the Victory of Adwa accessible, accurate, and alive for the next generation. This project turns technology into public-memory infrastructure—serving a visitor at the Adwa Victory Memorial Park today and growing into a national virtual heritage platform tomorrow. ## Status and vision The core MVP is built and operational: a mobile-first museum companion with a React experience, Fastify BFF, exhibit experiences, and containerised local development. We are opening the project to scale it into a resilient digital heritage and tourism platform for on-site discovery, remote exploration, and responsible preservation of community knowledge. ## Access architecture and cost resilience | Access path | Experience | Cost-resilience design | | --- | --- | --- | | **Primary: QR scanner** | Instant exhibit metadata lookup from a camera-scanned QR code. | The exhibit identifier resolves locally or from catalogued metadata; normal high-density traffic needs no AI call. | | **Photo scanner** | Artifact recognition when a QR code is unavailable. | Server-side perceptual hashing and cached matches avoid repeated Gemini Vision inference. | | **Virtual tours** | Remote 3D inspection and WebXR-ready walkthroughs. | Three.js and \`@google/model-viewer\` progressively load interactive models. | | **Offline-first fallback** | Essential guide functionality in poor-connectivity zones. | IndexedDB retains metadata and local assets; \`window.speechSynthesis\` provides speech when streaming audio or network access fails. | ## Technology stack - **Frontend:** React 18, Vite, Tailwind CSS, Zustand, Three.js, \`@google/model-viewer\`, GSAP. - **Backend BFF:** Node.js and Fastify, proxying Gemini Vision, Groq Llama/Whisper, and ElevenLabs TTS. - **RAG service:** Python 3.11, FastAPI, \`sente …