Nowadays, multimedia contents are widely used for various applications to ensure personal and public services. These contents are transferred through different kind of networks, from classical Adhoc networks to the recent Internet of Things (IoT). However, regardless the deployed network, the main goal remains the same: ensuring a better QoE and lower latency especially for emergency traffic. This domain attracts many research activities. In this paper, we tackle this challenging issue and we propose a cross-layer scheme to handle video streaming over wireless sensor networks. Our starting point is to provide a carrier sense aware disjoint multi-path routing protocol able to transmit efficiently video and regular data from many sources to a unique base station (BS). This protocol will cooperate with the application layer in order to provide a frame-aware solution to assign high priority to the most important frames and low priority for the least important ones. This priority concerns mainly the path that will be used by the frames and the queuing policy that will be applied. Our queuing policy involves enqueuing, de-queuing and congestion solving function in order to reduce video latency and enhance the reliability. Our scheme is explained through a theoretical study with illustrative examples. Many simulations have been conducted, they demonstrate improvements of some performance indicators such as packet data rate, delay, loss packet rate and user experience.