A Lightweight Video Streaming Framework for Low-End Systems Using FFMPEG and UDP for Real-Time Transmission

Real-time video transmission is crucial for applications such as face recognition systems, which demand continuous video input, processing, and output. This paper introduces a live streaming algorithm that relies solely on FFMPEG and UDP sockets to optimize CPU and memory efficiency. The method is structured into two phases: (1) Stream Generation, where video frames are captured, encoded (e.g., using H.264), and transmitted over UDP; and (2) Stream Reception, where incoming UDP packets are decoded and reconstructed into a coherent video stream. The proposed system supports both live CCTV feeds and pre-recorded video sources, offering low latency, high transmission stability, and efficient resource utilization. Experimental validation confirms the suitability of the approach for real-time video applications operating under resource constraints.