Current video encoding schemes used for ad hoc networks do not consider the characteristics of human visual system and the importance of image content when designing the encoder, so their encoding methods need to be improved. By analyzing motion information as well as the texture feature of video images, we propose a novel content-based video coding scheme C-FRMDC for robust video transmission over ad hoc networks. First determine the region of interest (ROI) of the video image before actual coding. Then divide the video sequence into ROI slice group and regular slice group using dynamic flexible macro block ordering (FMO) technology. After that, carry out multiple description coding scheme and redundant slices (RSs) technology to protect ROI slices. Finally transmit the bit-stream using multi-path transmission. The simulation results based on H.264/AVC show that our method improves the objective and subjective quality of the reconstructed video sequence obviously and ensure a robust video transmission over ad hoc networks.