How to differentiate audio and video data in Mdat in MP4 format

First in the Minf there is a VMHD and SMHD, then VMHD represents the video, SMHD represents the audio

And then in the Stsz,

Stsz Box
xx D8 size of stsz,20
7a:stsz
XX 00:version
XX 00:sample-size
XX 1C F1:sample-count
00 00 86 24 (starting from this, the size of the video data for each frame in the current chunk, which corresponds to offset, with offset to find the absolute address of the offset (in Mdat box), starting from the first byte of 000001b6, plus the corresponding number of video bytes, Is the size of this frame. A video chunk the first is a continuous video frame, then also contains other data, it is not clear that the data type behind the video data)
F9//Location of the first video frame
3B 04//Location of the second video frame
XX 1D F2