Abstract: Monitoring oil pipelines is crucial for effective infrastructure management and maintenance. It helps prevent threats such as vandalism and leaks, which can result in catastrophic events. Pipeline leaks pose significant environmental and economic risks, yet current detection methods are often expensive, slow, or unreliable, limiting their effectiveness for real-time applications. This research introduces a lightweight thermal imaging-based intelligent leak detection system that integrates Convolutional Neural Networks (CNNs), autoencoders, and knowledge distillation for deployment on edge devices. The proposed system addresses the challenges associated with existing pipeline detection techniques, such as large model sizes, high transmission latency, and excessive energy consumption. It utilizes thermal cameras to capture images of the pipeline, which are then compressed using an autoencoder. This compressed data is used to train a CNN model, which is further optimized through knowledge distillation. The model is trained and tested on real and synthetic data and deployed on a Raspberry Pi to simulate edge computing scenarios. Experimental results demonstrate improvement in detection accuracy, low inference latency, and an efficient transmission rate, confirming the system's suitability for real-time leak detection in remote and resource-constrained environments. This work contributes to the development of cost-effective, scalable, and energy-efficient solutions for pipeline monitoring.