Memora Vision
Memora Vision is an AI-powered visual memory assistant that transforms traditional CCTV and camera feeds into searchable, conversational intelligence. Instead of manually reviewing hours of footage, users can interact with the system using natural language queries such as “Where did I leave my backpack?”, “Who entered the room today?”, or “Summarize today’s activities.” Using multimodal AI models, computer vision, and event-based memory systems, Memora Vision understands scenes, tracks objects, detects events, and stores them as structured memories with timestamps. The system also supports real-time intelligent alerts, enabling users to create custom conditions like receiving notifications when someone arrives home or when unusual activity is detected. Built on AMD Instinct MI300X GPUs with ROCm acceleration, Memora Vision demonstrates how multimodal AI can turn passive surveillance systems into intelligent assistants capable of understanding, remembering, and responding to human environments in real time.
Category tags:
.png&w=96&q=75){kind=small}
.png&w=96&q=75){kind=small}
"This is a practical and useful application of multimodal AI for everyday life. Turning passive CCTV feeds into searchable, conversational intelligence solves real problems - manually reviewing hours of footage is tedious. The natural language query capability ("Where did I leave my backpack?") is compelling. Real-time alerts and event-based memory storage add practical value. Running on AMD MI300X shows good hardware utilization. Application of Technology: 🚀🚀🚀🚀🚀 5 - Uses multimodal AI models, computer vision, event-based memory systems. Understands scenes, tracks objects, detects events, stores memories with timestamps. Built on AMD Instinct MI300X GPUs with ROCm acceleration. Comprehensive technical implementation. Presentation: 🚀🚀🚀🚀 4 - Clear problem statement (manual footage review is tedious). Good explanation of features (natural language queries, real-time alerts, memory systems). Links to GitHub and demo provided. Well-structured. Business Value: 🚀🚀🚀🚀🚀 5 - Massive market. Every home and business with CCTV could use this. Security monitoring, finding lost items, activity summaries - all valuable use cases. Could be expanded to smart home integration. Originality: 🚀🚀🚀🚀🚀 5 - Very original. Turning CCTV into conversational intelligence is novel. The event-based memory system with timestamps is innovative. Natural language queries for video footage is a unique application. "
Sanem Avcil
For Innovators & Creators