Problems while running facedetect demo

Hi Erik,

We are looking at it.

Please bare with us until  we get back.

Regards,

Shankari.

Hi Shankari,

I had finally realised that the problem was caused by using a camera with PAL output format instead of NTSC.

Everything is working fine after modifying the code as suggested here:

http://e2e.ti.com/support/embedded/starterware/f/790/p/223776/810098.aspx

Regards,

Erik

Hello Shankari,

I was following your post about "Hardware connection details - Display - VGA monitor, video input - DVD player, Emulator - TI XDS560 Emulator"

Can you please suggest steps for TI XDS100v2 emulator for OMAP L138 board?

I was able to build the face detect but could not lunch debuger for my setup which is similar to your's above.

I really appreciate your help.

Thanks,

Roland

Hi Roland,

The steps will be the same except selecting the emulator type as " TI XDS100v2 emulator " in the target configuration.

Regards,

Shankari.

--------------------------------------------------------------------------------------------------------
Please click the Verify Answer button on this post if it answers your question.

--------------------------------------------------------------------------------------------------------

Hello Shankari,

Thanks for the information. I did that and it works. The other issue is I cannot see any rectangle on my face when the program runs.

Any little suggestion would be very appreciated.

Thanks,

Roland

Hello Shankari,

Thank you for the information. What really makes it work was waiting for about 3 minutes at same location. It seems the algo get confused when you move around and it has to restart the comparison from the beginnih. Well, with patience, the rectangle finally appeared. It worth using another person as target in the video while you go back and forth on implementation. 

Thanks,

Roland

Hi Shankari,

I have replaced the PAL camera with a OV7962 sensor connected through the 36 way camera connector on the LCDK. I'm having issues when trying to acquire images from this sensor because the VPIF Channel0 interruption is not being triggered.

What I've done is the following:

1) Supply the OV sensor through 1, 2 and 3 pins in J16 connector.

2) Generate an external 24MHz clock and fed it trough the sensor into the LCDK using the CAM_CLK pin (no. 9) in J16.

3) Connect VSYNC, HSYNC and data bit signals in 11 to 24 in J16. I have measured with an oscilloscope VSYNC and HSYNC in the connector pins and seem to be OK (see images, VSYNC below, HSYNC above).

These signals doesn't fit exactly the ones shown in page 1753 in spruh77a document, where it seems that both VSYNC and HSYNC signals have to be high to capture data. Because of this I have modified the bit 14 in C0CTRL register in order to invert the vertical pixel valid signal inseide the VPIF, by adding the correspodning code line in SetUpVPIFRx function in facedetect_lcdk.c.

4) Code modifications. Apart from the previous one I have also included some other modifications in SetUpVPIFRx .

/*    // Interrupt after capturing the bottom field of every frame
    VPIFCaptureIntframeConfig(SOC_VPIF_0_REGS, VPIF_CHANNEL_0, VPIF_FRAME_INTERRUPT_BOTTOM);
    // Y/C interleaved capture over 8-bit bus
    VPIFCaptureYcmuxModeSelect(SOC_VPIF_0_REGS, VPIF_CHANNEL_0, VPIF_YC_MUXED);
    // Capturing 480I (SD NTSC)
    //VPIFCaptureModeConfig(SOC_VPIF_0_REGS, VPIF_480I, VPIF_CHANNEL_0, 0, (struct vbufParam *) 0);
    VPIFCaptureModeConfig(SOC_VPIF_0_REGS, VPIF_PAL, VPIF_CHANNEL_0, 0, (struct vbufParam *) 0);
*/
    VPIFCaptureRawIntlineConfig(SOC_VPIF_0_REGS, 640*480);
    VPIFCaptureRawVvinvSet(SOC_VPIF_0_REGS, VPIF_RAW_VVALID_INVERT);
    VPIFCaptureFieldframeModeSelect(SOC_VPIF_0_REGS, VPIF_FRAME_BASED);
    VPIFCaptureIntrprogModeSelect(SOC_VPIF_0_REGS, VPIF_CHANNEL_0, VPIF_CAPTURE_PROGRESSIVE);
    VPIFCaptureIntframeConfig(SOC_VPIF_0_REGS, VPIF_CHANNEL_0, VPIF_FRAME_INTERRUPT_TOP);
    VPIFCaptureCapmodeModeSelect(SOC_VPIF_0_REGS, VPIF_CHANNEL_0, VPIF_CAPTURE_RAW);

As I mentioned, I don't receive any image and when I put a breakpoint in VPIFIsr function it never reaches there, so the interrupt is not being triggered. Any hint about what can be happening and how could I solve this issue would be very grateful.

Best regards,

Erik

Hi again,

I've managed to modify sensor's output to generate an inverted VSYNC signal. Now the shape of the signal matches the expected one but the interrupt is not triggered.

I've seen in the documentation that the expected vertical blanking time is at least 3 lines. The signal that I'm getting has only two lines width. Might this be the problem? Is it possible to reconfigure the VPIF to accept 2 lines vertical blanking time?

Regards,

Erik

Hi,

Anyone of the TI's support team could help me? I'm stucked with this issue and I need to go on as fast as possible.

Regards,

Erik

up

up

Hello Erik,

I am working on raw capture with OMAP-L138 LCDK too.

I think that you might have misinterpret the second parameter in the following function:

VPIFCaptureRawIntlineConfig(SOC_VPIF_0_REGS, 28);

The second parameter indicates the number of lines (hexadecimal).

I am not sure about VPIF_Rx configuration that I have made because I am having  some issues with i2c  and I cannot check it,so my obove suggestion might be wrong.

Finally,I have a question regarding your original post.You mention that you generated a 24 MHz clock , did you do  that with CDCE913Init() function?

Regards,

Katerina

Hello Katerina,

Thanks for your input. You are right, I might have misinterpreted this parameter, but I have just tested different options for this parameter and in any case I am not able to get an interruption from the VPIF port, thus no image read.

The external clock that I have used is not generated with the CDCE913 but a clock generated externally with an specific HW that we have in the OmniVision sensor’s board, then we feed this clock into the LCDK through pin 9 in J16.

Regards,

Erik

Hello Erik,

thank you for the clarification about the clocking!As for the VPIF interrupt triggering I have no further suggestions!

Regards,

Katerina

hi Erik

i also have OMAPL138LCDK board  i want to run the facedtect demo i want to know which camera u have used for this purpose  i am using XDS100V2  emulator and i have ccs5.5 and where from u have got the coding for interfacing of camera and the image processing done on it please provide me the details 

thank you

Hello !!

I have the some problem I modified the program as suggested 

http://e2e.ti.com/support/embedded/starterware/f/790/p/223776/810098.aspx

But I have any thing in the screen.

Could you please help me or attach the project that work.

Think you

Hi Eric

Have you solved your problem above? I face the same problem now and look for some help.

I'll be grateful if you reply.

Shreek