Build a React PWA called 'Leaf Snap' using WebInfer with vision capabilities. UI: camera for capture, identification result card, my plants collection, nearby sightings map. Capture: photograph plant (leaf, flower, or whole plant). Multiple shots supported for better accuracy. Vision model analysis with generateObject returns { species: string, commonNames: string[], family: string, confidence: number, alternativeMatches: [{ species: string, confidence: number, differentiatingFeatures: string }], identifyingFeatures: string[] }. Info generation: use generateObject to return { description: string, nativeRegion: string, habitat: string, toxicity: { humans: 'safe'|'mildly-toxic'|'toxic'|'deadly', pets: 'safe'|'toxic-dogs'|'toxic-cats'|'toxic-all', notes: string }, care: { light: string, water: string, soil: string, temperature: string, humidity: string, fertilizer: string }, commonProblems: string[], propagation: string, growingDifficulty: 'easy'|'moderate'|'expert', edibility?: string, similarSpecies: [{ species: string, warning: string }] }. PROMINENT toxicity warning if toxic to pets. Save to collection: track your plants with photos, care schedules. Location tagging via Geolocation: where you spotted wild plants. Offline: core identification works offline. Watering reminder integration: set care schedules. Affiliate: links to buy seeds/plants, care products.