Linbian Usage Guide

Terminal

On Linbian OS, the default terminal application is LXTerminal.

You can open it from Application Menu -> System Tools.

Camera Module

download source code from github

• ldcam developed base on gstreamer

git clone https://github.com/lindenis-org/ldcam-gst.git

compile

• compile on v5 board===

you can execute "make" directly

• compile on your own env

you should create the Makefile by yourself

user commd

you can use "./ldcam -?" or "./ldcam --help" to get the cmd instructions

Gstreamer

Camera source

• lindeniv4l2src: Video (video4linux2) Source
  

Use for CSI Camera
Example:
sudo gst-launch-1.0 lindeniv4l2src device=/dev/video1 ! video/x-raw,width=1920,height=1080,format=I420 ! videoconvert ! sunxifbsink video-memory=24

Omx hardware H.264 decoder and encoder

• omxh264dec: OpenMAX H.264 Video Decoder
  

Example:
sudo gst-launch-1.0 filesrc location=1c.mp4 ! qtdemux name=demux demux.video_0 ! queue ! h264parse ! omxh264dec ! videoconvert ! sunxifbsink

• omxh264videoenc: OpenMAX H.264 Video Encoder
  

Example:
sudo gst-launch-1.0 lindeniv4l2src device=/dev/video1 ! video/x-raw,width=1920,height=1080,format=I420 ! videoconvert ! omxh264videoenc target-bitrate=5000000 ! mpegtsmux ! filesink location=1.ts

Display sink

• sunxifbsink: Accelerated console framebuffer video sink for sunxi-based devices
  

Example:
sudo gst-launch-1.0 -v videotestsrc pattern=ball name=left ! sunxifbsink

MPP

MPP is the media process platform of Allwinner V5, including following components.

• system control, API family: AW_MPI_SYS

This component deal with the media hardware initialization, other components initialization and component state management and etc.

• video input, API family: AW_MPI_VI
• video output, API family: AW_MPI_VO
• image stitch engine, API family: AW_MPI_ISE
• video encode, API family: AW_MPI_VENC
• video decode, API family: AW_MPI_VDEC
• mux, API family: AW_MPI_MUX
• demux, API family: AW_MPI_DEMUX
• audio input, API family: AW_MPI_AI
• audio output, API family: AW_MPI_AO
• audio endode, API family: AW_MPI_AENC
• audio output, API family: AW_MPI_AOUT
• Region, API family: AW_MPI_RGN

Allwinner official documents

• MPP Application Development Reference Manual, Chinese version, English version.
• ISP Module User Guide, Chinese version, English version.
• EVE Application Development Reference Manual, Chinese version, English version.
• Human Counting Application Development Reference Manual, Chinese version
• Motion Detection Application Development Reference Manual, Chinese version
• Binocular depth of field (DOF) Application Development Reference Manual, Chinese version

Samples

Please refer to Demo programs.

Demo programs

Allwinner release a series of MPP samples. You can download from Github.

$ git clone https://github.com/lindenis-org/mpp_sample ~/mpp_sample

After downloading, You will get the following directory.

app/
Makefile.param
README.md
sample_adec/
sample_aenc/
sample_ai/
......

There is a instruction name readme.txt in every sample, you should read the document first when you running the sample

And then you can know how to config the sample, some sample needed be configured,maybe config file is sample_xxx.conf.

app/ includes a set of MPP samples with a command line based menu. sample_xxx/ is a separate sample of the scene. Root privileges is required when run these sample. e.g.

$ sudo ./app/release/app_mpp_sample

***************************************************************
  1 : vi+(venc+rtsp)+vo
  2 : vi+(venc+rtsp)
  3 : vi+vo
  4 : vi+ise+(venc+rtsp)+vo
  5 : [vi+venc+vo]+[ai+aenc+ao]+rtsp
  6 : vi->yuv+venc+rtsp
  7 : vi+venc+mux
  8 : vi+(venc+mux)+vo
  9 : demux+vdec+vo
 10 : ai+aenc+mux+ao
 11 : demux+adec+ao
 12 : Quit
***************************************************************
 Please choice 1~12 num:

RTSP

Location

app_mpp_sample

Build

$ cd ~/mpp_sample/app
$ make

Running

$ sudo ./release/app_mpp_sample

There will shows the following message when you execultive the cmd, and type the num according the menu.

***************************************************************
  1 : vi+(venc+rtsp)+vo
  2 : vi+(venc+rtsp)
  3 : vi+vo
  4 : vi+ise+(venc+rtsp)+vo
  5 : [vi+venc+vo]+[ai+aenc+ao]+rtsp
  6 : vi->yuv+venc+rtsp
  7 : vi+venc+mux
  8 : vi+(venc+mux)+vo
  9 : demux+vdec+vo
 10 : ai+aenc+mux+ao
 11 : demux+adec+ao
 12 : Quit
***************************************************************
 Please choice 1~12 num: 1

***************************************************************
  1 : Set VI+VENC scene     (Resolution,    default:4K@25fps+720P@25fps)
  2 : Set VENC Payload Type (H264/H265,     default:H264)
  3 : Set VENC RC Mode      (CBR/VBR/FIXQP, default:CBR)
  4 : Set VENC Profile      (BL/MP/HP,      default:Main Profile)
  5 : Set VENC rotate       (0/90/180/270,  default:rotate_0)
  6 : Save confige and run this sample
***************************************************************
 Please choice 1~6 num: 1

 ***************** Choice VI+VENC scene **************************
 [0]:VI_4K@30fps + VENC(4K@30fps+VGA@30fps)
 [1]:VI_4K@25fps + VENC(4K@25fps+720P@25fps) [default]
 [2]:VI_2K@30fps + VENC(2K@30fps+720P@30fps)
 [3]:VI_1080P@30fps + VENC(1080P@30fps+720P@30fps)
 [4]:VI_2880x2160@30fps + VENC(2880x2160@30fps+1080P@30fps)
 [5]:VI_2592x1944@30fps + VENC(2592x1944@30fps+1080P@30fps)
 Please choose VI+VENC scene ID 0~3 or (q-Quit): 1

 ***************************************************************
  1 : Set VI+VENC scene     (Resolution,    default:4K@25fps+720P@25fps)
  2 : Set VENC Payload Type (H264/H265,     default:H264)
  3 : Set VENC RC Mode      (CBR/VBR/FIXQP, default:CBR)
  4 : Set VENC Profile      (BL/MP/HP,      default:Main Profile)
  5 : Set VENC rotate       (0/90/180/270,  default:rotate_0)
  6 : Save confige and run this sample
***************************************************************
 Please choice 1~6 num: 6

Test result

• You can see that the two remote device can get the data at the same time

• There are two output data channel, you can get 720p from channel1 and 4K data from channel0

• Tt also can be viewed by mobile phone, a simple IPC mode.

Dual Camera

Location

sample_virvi2ise2vo

Build

$ cd ~/mpp_sample/sample_virvi2ise2vo
$ make

Running

$ sudo ./sample_virvi2ise2vo -path sample_virvi2ise2vo.conf

Test result

• Two ISP can work at the same time

• And each one can send 4K 30fps data out when you preview the picture

• Device can support 4K 15fps encode when two ISP work at the same time

Face detection

Face detection based on EVE.

Location

sample_face_detect

Build

$ cd ~/mpp_sample/sample_face_detect
$ make

Running

$ sudo ./sample_face_detect

The picture of face crop and store at ./pic directory when detected.

Test result

• Support face detect which the data source minmum pixels is 20

• Support face detection and tracing at the same time, the maximum number is 16

• The maximum detect face number is 128

Motion detection

Motion detection based on CVE.

Location

sample_AILib/sample_vi2MOD

Build

$ cd ~/mpp_sample/sample_AILib/sample_vi2MOD
$ make 

Running

$ sudo ./sample_vi2MOD -path sample_vi2MOD.conf

The information of detected target was stored at ./DTCAResult.bin .

Human counting

Hunman counting based on CVE.

Location

sample_AILib/sample_vi2HCNT

Build

$ cd ~/mpp_sample/sample_AILib/sample_vi2HCNT
$ make

Running

$ sudo ./sample_vi2HCNT -path sample_vi2HCNT.conf

The information of detected target was stored at ./HCNTResult.bin .

VLPR

VLPR based on CVE.

Location

sample_AILib/sample_vi2VLPR

Build

$ cd ~/mpp_sample/sample_AILib/sample_vi2VLPR
$ make

Running

$ sudo ./sample_vi2VLPR -path sample_vi2VLPR.conf

The information of detected vehicle license plate was stored at ./VLPRResult.bin .

Binocular depth information image

BDII based on CVE.

Location

sample_AILib/sample_BDII

Build

$ cd ~/mpp_sample/sample_AILib/sample_vi2BDII
$ make

Running

$ sudo ./sample_vi2BDII -path sample_vi2BDII.conf

The output is a 8bit-gray-scale picture, stored at ./BDIIResult.bin .

OpenCV

sample_virvi2opencv2vo

OpenCV version 4.0.1
This is a simple sample for motion detection via opencv.
When running this sample, if the picture captured by the camera is still, it will be black screen. If there is a moving object, the outline of the moving object will be displayed. When the mean difference of the real-time image frame difference is greater than the sensitivity value, the following printing appears, indicating that there is a significant change in the image or something is moving.

Something is moving! mean_diff:10
Something is moving! mean_diff:70
Something is moving! mean_diff:94
Something is moving! mean_diff:120
Something is moving! mean_diff:362
Something is moving! mean_diff:34
Something is moving! mean_diff:30

Location

sample_virvi2opencv2vo

Build

$ cd ~/mpp_sample/sample_virvi2opencv2vo
$ make

Running

$ sudo ./sample_virvi2opencv2vo -path ./sample_virvi2opencv2vo.conf

You can change the conf in the sample_virvi2opencv2vo.conf

1. auto_test_count : test times
2. get_frame_count : number of frame in one time test
3. dev number: video node
4. src_width * src_height:720p/1080p;
5. src_frame_rate: 25
6. pic_format is nv21
7. enable_disp_color: 0 disable, 1 enable
8. moving_detect_sensitivity >=1 , 1 is most sensitive

Test result