Edge AI: Deploying Machine Learning Models on Resource-Constrained Devices

Modugu Dileep Kumar1, Sangoju Venkata Sri Lakshmi Sohini 2
1Assistant Professor, Department of Computer Science and Engineering, St. Martin’s Engineering College
2UG Student, Department of Computer Science and Engineering, St. Martin’s Engineering College
Email: dilee1213@gmail.com1, sangojuvenkatasrilakshmisohini@gmail.com2

Abstract:Edge AI focuses on deploying machine learning models directly on resource-constrained devices such as smartphones, IoT sensors, and embedded systems. This project explores efficient techniques for running intelligent models locally without relying on cloud infrastructure. The main objective is to reduce latency, improve data privacy, and enable real-time decision making. Lightweight models and optimization methods like pruning, quantization, and knowledge distillation are utilized to fit limited computational resources. The project involves selecting suitable machine learning algorithms and converting them into optimized formats for edge deployment. Frameworks such as TensorFlow Lite and ONNX Runtime are used to support model execution on low-power devices. Performance is evaluated based on accuracy, inference speed, memory usage, and power consumption. The system demonstrates how edge devices can perform tasks like image classification, object detection, or anomaly detection efficiently. Challenges such as limited memory, processing power, and energy constraints are addressedthrough model optimization strategies. The project also highlights trade-offs between model complexity and performance. Real-world use cases include smart surveillance, healthcare monitoring, and industrial automation. By shifting computation closer to data sources, network dependency is minimized. The approach ensures faster responses and enhanced security by keeping sensitive data on-device. Overall, this project showcases the potential of Edge AI in building scalable and efficient intelligent systems.