Hi,

I bought a V5 to stream 4K from the camera over my network. I have the sample working but I wonder what the correct command line would be for a 4K RTSP stream?


I have been trying to get it working with this line:

Hi,

Does anybody already used CSI1 as a BT.1120 16bits input ? Would you mind to share the configuration ? My input signal is a 1080p60 4:2:2 and I can not figure out how to configure the CSI1.

Best.

hello~

I want test extended camera via BT112 CSI line.
but I couldn't find suitable spec for CSI male connector on bottom side.

Could you let me know suitable male connector for this or show me current female connector data sheet?
   

These AI kernels would run 50% faster if I were able to compile them with gnueabihf calling conventions.  It is a major pain to mix gnueabi and gnueabihf on the same system. Life would be far simpler if AW would provide a gnueabihf version of their libraries.

A huge downside to mixing gnueabi and gnueabihf is that you end up with two copies of all of the system libraries in memory at the same time. For example you have to have two versions of the C run-time, etc, etc.

I've done more benchmarking and it appears that not having gnueabihf hurts AI core performance a lot. On large models like inception it is 50%. On a small model like squeezenet it is 300%.  Not having hf is preventing register passing via the FPU registers.

Rasp3b has same quad core Cortex-A7, armv7l as the V5 does. I should be able to get the same benchmark results or very close on the V5. Only significant difference is gnueabihf on Rasp3b.

squeezenet_v1.1  
    v5 298ms rasp3b 94ms
inception_v3
    v5 3592ms rap3b 2074ms

Two messages repeatedly come out to my console while I am doing DTCA processing (AW_AI_CVE_DTCA_Process). They are:

  'Fininished'

and 

  'stImage.s32Width = 640,stImage.s32Height = 360'


Is there some way to turn them off?

I have the output of a video encoder and the output of an audio encoder bound to a muxer group and channel.  The video is 1080p@30fps compressed to h264.  The audio is 16bit, 48K audio compressed with AAC.  I have asked the muxer to write MPEG-TS files.

I am manually gathering the video (AW_MPI_ISE_GetData) and the audio (AW_MPI_AI_GetFrame) in my own threads and sending them into the encoders in the normal way (AW_MPI_VENC_SendFrame & AW_MPI_AENC_SendFrame).

The audio in the resulting MPEG-TS files seems to be 'behind' the video by several hundred milliseconds.  Voices, for instance, don't sync with the lip movements in the video.  How can I fix this?

I noticed that the timestamp on the audio data (AUDIO_FRAME_S mTimeStamp) is set to zero when I retrieve it with AW_MPI_AI_GetFrame.  The video PTS (VIDEO_FRAME_S mpts) does seem to be set correctly.  My thought was that I needed to sync the video and audio timestamps (by using clock()) so the muxer can sync them up.  However, when I set mTimeStamp of the audio frames to something non-zero, I get the follwing error message when I call AW_MPI_AENC_SendFrame:

E1202 19:23:16.867974  1173 AudioEnc_Component.c:971]       <AudioEncEmptyThisBuffer>     fatal error! RecRender should not call this function!

The error return from AS_MPI_AENC_SendFrame is '-1'.  

Am I supposed to set the audio timestamp?  If not, how can I get the audio and video to sync up?

One way to eliminate the need for specialized tools to program the EVE AI hardware would be to provide a tensorflow driver for it. Then you could use the tensorflow tools to make the classifers and run them on the V5.

In our use case we'd like to make classifiers to detect audio patterns.

This Qualcomm chip is similar to the V5 and it uses tensorflow...
https://www.cnx-software.com/2018/04/12/...lications/

The mpp samples do not run on the middleware from this commit "eagle: update to support isp tuning tool".

I have verified this by backing up to the previous version of the middleware and checking that the samples do run.

We need the matching version of the samples that goes with the isp tuning tool version of the middleware.

Did you guys build this, or is it copied from somewhere?

Basic rootfs
Download the basic rootfs from FTP server with user 'ftp' (password: lindeni) and store to debian directory.

• stretch-alip.tar.gz
  

Has anything been done to integrate the V5 AI hardware with OpenCV?