In this paper, a remote video conferencing system based on cloud video is proposed and implemented. The system uses a mesh transfer card for MV-440 single channel high definition MPEG4 images. This paper introduces the design and implementation of online video system based on DELPHI+ VDK. The main functions are high-definition video, camera control, data processing, image adjustment, sound Settings and different software styles. This method fixes images that have been damaged on the network by resending and FEC encoding. The image quality has been greatly improved in the unsatisfactory network environment. The experiment proves that the multi-person online intelligent network training system has the characteristics of small load, good controllability and remarkable management effect.