Decode RTP H264 Encoded Video with libimxvpuapi

hsuq7052 — Tue, 05 Mar 2024 07:24:03 GMT

I'm working with an imx8mp and using a generated Yocto Image based off 5.15-kirkstone and a kernel version version 5.15.52+g244c28a772f6. .I'm trying to decode a H.264 RTP video stream, I'm removing the 12 bytes from the RTP header and calling imx_vpu_api_dec_push_encoded_frame with the encoded frame data being the Fragmented Frame (FU indicator + FU header + FU payload). I'm getting an error from the VPU, as it never has a frame to decode and continuously outputs codec state CODEC_NEED_MORE even after pushing frames in successfully. I'm also uncertain how the VPU is acquiring the data from SPS and PPS NALUs, I see a reference in the decode-example but not any mention to how the VPU lib is supposed to get that info. I've also tried to depacketize the frames into whole frames but then I get an error `nal_ref_idc cannot be null`. Am I missing something or the vpu lib not able to decode this type of H.264 video stream?

Libimxvpuapi Version: 2.2.1
Platform: IMX8MP

topic Decode RTP H264 Encoded Video with libimxvpuapi in i.MX Processors

Decode RTP H264 Encoded Video with libimxvpuapi

Re: Decode RTP H264 Encoded Video with libimxvpuapi