SteelCount Pro: AI-Powered Steel Rod Detection and Inventory Management System Using Computer Vision and Progressive Web Application

Steel rod inventory management at construction sites and warehouses relies on manual counting methods that are time-consuming (30–60 minutes per bundle) and error-prone (15–20% miscounts), leading to significant financial losses in the Indian construction industry. This paper presents SteelCount Pro, an AI-powered Progressive Web Application that automates steel rod detection, counting, size classification, and cost estimation from a single photograph using computer vision. The system employs a Roboflow-trained object detection model (Version 12) achieving 85–99% detection accuracy across eight IS 1786 standard rod sizes (6mm to 32mm). A novel inventory calculation pipeline automatically classifies detected rods by diameter, computes individual and total weights using the IS 1786 standard formula W = D²/162 × L, and generates instant cost estimates using configurable per-size pricing. The application supports two input modes (live camera capture and gallery upload), maintains a local scan history of up to 50 records with JSON export capability, and operates in an offline demo mode for environments with limited connectivity. Built with React 19 and TypeScript as a PWA with Capacitor for native Android/iOS deployment, the system reduces counting time by 90% and virtually eliminates human counting errors. Validated through 18 functional test cases, the platform provides a practical, zero-cost solution for construction inventory management.