AI-Based Object Measurement Using MiDaS Depth Estimation and MobileNetV3 Edge Detection

In the modern age, Artificial Intelligence (AI) is transforming industries with solutions to functions based on human estimation and manual equipment conventionally. Among them, object dimension measurement, essential across applications from health and biometrics to industrial quality control, is a major area of application. In this research, an AI system for estimating real-world object length based on depth estimation, edge detection models, and computer vision is investigated. The computer vision problem was created in two stages and initially provides a Streamlit-based prototype coded in Python through which users may upload pictures or provide live camera shots to measure objects by choosing two points manually. The MiDaS model for monocular depth estimation and MobileNetV3 for edge detection are used to assist with precise measurement without physical reference objects. Measure results are reported in measurements such as meters or centimeters with visual overlays for ease of use. The second is creating an end-to-end React web application for mobile-first use. The build contains ONNX-converted AI models and a minimal UI, live camera streaming, live video point tagging, and page modules such as Home, Help, and Settings. With onboarding-type design and instructional popups integrated, the app is able to sustain a user-friendly experience without compromising measurement precision. This work enriches the AI-based measurement literature in medical imaging, biometrics, and industrial measurement with a web-based, flexible solution alternative to domain-specific solutions. Using lightweight AI models with responsive UI promises scalable, user-friendly measurement solutions deployable across devices and environments.