To build an AI-powered retail intelligence platform that delivers real-time shopper analytics, automated inventory visibility, and proactive queue management through on-device AI,enabling retailers to reduce stock-outs, improve customer experience, optimize staffing, and increase operational efficiency while maintaining privacy and minimizing cloud dependency.
“We are building an intelligent, offline-capable retail camera system that watches store aisles and checkout lines locally to prevent empty shelves and long queues. By processing video right on the device without sending data to the cloud, it protects customer privacy and works seamlessly even in stores with poor internet.”
Deconstructed Problem Blueprint
WHAT TO BUILD?
An edge-AI-powered retail intelligence platform combining computer vision models running locally on edge hardware with a central store-operations dashboard.
WHO IS IT FOR?
Retail store owners, supermarket managers, and staff who struggle with manual inventory checks and long billing lines, as well as shoppers who want faster checkouts and fully stocked shelves.
WHY DOES IT EXIST?
Traditional retail surveillance requires heavy cloud bandwidth, invades customer privacy, and fails completely during internet outages. Furthermore, manual inventory tracking leads to frequent stock-outs and lost revenue in high-volume retail environments.
INPUT → PROCESS → OUTPUT ARCHITECTURE PIPELINE
RTSP video streams from store security and shelf-facing cameras, local configuration parameters, and basic product planograms.
On-device object detection and tracking for footfall and dwell time, shelf-image comparison for out-of-stock detection, queue-length estimation via spatial heuristics, and anonymous re-identification.
Real-time staff alerts via webhooks/SMS, interactive heatmaps, queue congestion predictions, and a comprehensive manager dashboard showing KPIs.
CONSTRAINTS
- Must run completely offline without relying on continuous cloud connectivity.
- Must preserve customer privacy by avoiding facial recognition or PII storage.
- Must execute real-time inference on low-power edge hardware with minimal latency.
- Must integrate smoothly with existing retail POS and inventory systems.
JURY METRICS
- Edge inference latency (frames per second)
- Precision and recall for out-of-stock and low-stock detection
- Queue congestion prediction accuracy
- Bandwidth consumption reduction compared to cloud-streaming solutions
- Dashboard UX simplicity and responsiveness
CORE BOTTLENECK
Optimizing multiple concurrent computer vision models (object detection, shelf compliance, and queue tracking) to run smoothly in real-time on resource-constrained edge hardware without overheating or lagging.
Why You Should Choose This PS (Pros)
- Sponsored by Qualcomm, giving teams a massive edge if they target Qualcomm Snapdragon/RB series hardware.
- Extremely high commercial value with clear product-market fit for modern supermarkets and kirana stores.
- Visually appealing demo potential with live heatmaps and bounding boxes.
Challenges & Risks to Watch (Cons)
- Requires optimizing heavy deep learning models for edge devices (quantization, pruning).
- Simulating multiple physical camera angles and shelf states during a hackathon can be tedious.
- Crowded domain if teams only build a basic CCTV object counter.
Recommended Tech Stack & Grand Finale Features
Killer Features to Impress the SIH Jury
View Official SIH 2026 Portal Raw Text▼
Background India's retail sector includes millions of neighborhood stores, supermarkets,pharmacies, and large-format retail outlets that serve high customer volumes every day. Retailers face challenges such as inventory shrinkage, stock-outs, long billing queues, inefficient shelf replenishment, and limited visibility into shopper behavior. Many stores, especially in Tier-2 and Tier-3 cities, also operate with constrained internet connectivity and require solutions that can function reliably without continuous cloud access.
Recent advances in edge AI allow cameras and sensors to perform real-time analytics directly on local devices, enabling faster decisions, improved privacy,reduced bandwidth consumption, and uninterrupted operation even during connectivity outages. Hybrid and edge AI approaches are increasingly being adopted for real-time monitoring and decision support across multiple industries.
Description Design an Intelligent Retail Analytics System that uses smart cameras and on-device AI to monitor retail operations in real time. The system should analyze shopper movement, inventory levels, and checkout queues without requiring constant cloud processing.The solution should automatically identify customer traffic patterns, measure dwell time in different store sections, detect out-of-stock products, monitor shelf compliance, and predict queue congestion before it impacts customer experience.AI inference should happen locally on the edge devices to enable low-latency decisions while preserving customer privacy and minimizing network dependency.The system should convert video streams into actionable business insights that help retailers improve operational efficiency, optimize staffing, increase product availability, and enhance customer satisfaction. Edge-based analytics can provide real-time intelligence while reducing dependence on cloud connectivity.
Expected Solution The proposed solution should implement some or all of the following: