HTTP Adaptive Streaming is a successful and largely adopted content delivery technology. Yet poor bandwidth prediction, notably in mobile networks, may cause bit-rate oscillations, increased segment delivery delays, video freezes, and may thus negatively impact the end user quality of experience. To address this issue, we propose to exploit the stream prioritization and termination features of the HTTP/2 protocol to achieve video frame filtering and scheduling, so as to maximize the amount of video data received on time by the client. We evaluate with optimal scheduling and filtering algorithms the maximum gain we may expect from such delivery schemes where video frames are carried in dedicated HTTP/2 streams. Evaluation shows that our HTTP/2-based video frame scheduling scheme brings benefits for video quality.