Defense Surveillance system uses the deep learning methods in addition with the ESp32 module to detect humans and weapons in real time. The YOLOv8(You Only Look Once) algorithm is used to find the weapons and human using video capturing techniques and the system latency is been reduced with the feedback video signal from the ESP32. The deep learning models are developed and trained in Visual Studio Code with the aid of frameworks like YOLO and OpenCV. To improve the detection accuracy the system also integrates sensor data in difficult scenarios, such as low light or camouflage. In defense operations, this real-time AI-powered solution facilitates quick decision-making and improves situational awareness. The system’s efficiency and portability are guaranteed by the use of a lightweight, inexpensive ESP32 module, which makes it appropriate for broad deployment in defense settings. The proposed method Yolov8 is compared with SSD and Mask R CNN and calculated the performance metrics like accuracy, Precision and recall.