The increase in the frequency and intensity of disasters has created an urgent need for efficient and cost-effective emergency response systems. Traditional surveillance methods, such as helicopters, are often limited by their high operational costs, inability to operate effectively in hazardous environments, and significant risks to human safety, especially during mechanical failures or adverse weather conditions. This paper presents SAVER (Surveillance and Aerial Visual Emergency Response), a Unmanned Aerial Vehicle (UAV)-based system designed to enhance disaster management through advanced object detection and thermal mapping. The system employs the YOLOv5m6u model for real-time object recognition, achieving a precision of 0.78, recall of 0.84, and an F1 score of 0.81 over 50 epochs of training using a custom-labeled dataset of 759 images. The proposed model offers dual-mode data transfer (Wi-Fi-based live streaming and onboard storage), enabling reliable data capture even in areas with limited internet connectivity. In addition, SAVER integrates two-way audio communication, allowing real-time interaction between UAV operators and affected individuals, and employs crisis impact classification to prioritize resource allocation based on detected human density. The solution provides cost-effective thermal mapping without the use of expensive sensors, significantly reducing operational costs compared to traditional methods. Field testing demonstrated SAVER’s capability to detect objects with an average frame processing time of 25 ms, enhancing response efficiency across diverse disaster scenarios, including urban flooding, train accidents, and traffic control. Future improvements include multi-drone coordination and shortest path algorithms to further optimize emergency operations.