cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Highlighted
Observer
Observer
358 Views
Registered: ‎04-23-2020

FPS question on ZCU102

Jump to solution

Hi

I try to run ZCU102 sample adas_detection.It displayed FPS on the video.

But I comfused about FPS variety.

FPS started about 14 and improve to 35 gradually and show below error.

擷取3.JPG

I think FPS should be stable.

Is it true or I loss some step to cause error?

 

yumei

0 Kudos
1 Solution

Accepted Solutions
Highlighted
Adventurer
Adventurer
288 Views
Registered: ‎07-20-2017

@yu_mei You can use any third party software to convert the encoded video into raw format. Then use that to test the model. 

DPU cannot be used to resize the video. You will have to either do that in PS(CPU) before passing the input to DPU or in PL(FPGA) for more acceleration.

There are also pruning techniques you can use to compress the neural network model for faster performance.

Regards, abhidan@logictronix.com
Please mark the Answer as "Accept as solution" if information provided solves your query. Give Kudos if you think it was helpful and reply oriented.

View solution in original post

7 Replies
Highlighted
Adventurer
Adventurer
337 Views
Registered: ‎07-20-2017

@yu_mei 

This thread has the similar issue: https://forums.xilinx.com/t5/AI-and-Vitis-AI/DNNDK-quot-dblind-warning-quot-Error-retrieving-accessibility/td-p/1057686

FPS can also depend on the encoding of the video used or number of the detected objects on the video frame.

Regards, abhidan@logictronix.com
Please mark the Answer as "Accept as solution" if information provided solves your query. Give Kudos if you think it was helpful and reply oriented.
0 Kudos
Highlighted
Observer
Observer
298 Views
Registered: ‎04-23-2020

@idiotic_genius 

Hi

Do you have any suggestion about encoding of the video to improve performance? 

Can I use dpu to resize video?

Thank you

 

yumei

 

Tags (1)
0 Kudos
Highlighted
Adventurer
Adventurer
289 Views
Registered: ‎07-20-2017

@yu_mei You can use any third party software to convert the encoded video into raw format. Then use that to test the model. 

DPU cannot be used to resize the video. You will have to either do that in PS(CPU) before passing the input to DPU or in PL(FPGA) for more acceleration.

There are also pruning techniques you can use to compress the neural network model for faster performance.

Regards, abhidan@logictronix.com
Please mark the Answer as "Accept as solution" if information provided solves your query. Give Kudos if you think it was helpful and reply oriented.

View solution in original post

Highlighted
Explorer
Explorer
235 Views
Registered: ‎01-20-2019

Hii @idiotic_genius 

It would be great if you can mention some of the useful tool links to convert the encoded video file to the raw format and also how do we can read the video frames from this raw video file?

Thank you in advance.

0 Kudos
Highlighted
Adventurer
Adventurer
197 Views
Registered: ‎07-20-2017

You can use FFMPEG tool. 

https://ffmpeg.org/

You can use OpenCV library functions to read the video frames from this raw video files.

Regards, abhidan@logictronix.com
Please mark the Answer as "Accept as solution" if information provided solves your query. Give Kudos if you think it was helpful and reply oriented.
0 Kudos
Highlighted
Explorer
Explorer
167 Views
Registered: ‎01-20-2019

@idiotic_genius 

Thank you for your quick response.

I have decompressed .mp4 by running the below command.

ffmpeg -v error -i input.mp4 -pix_fmt bgr24 -f rawvideo -an -sn output.bgr24

Below is my OpenCV program to read the frame from output.bgr24 file.

fstream infile;
unsigned int i=1;
infile.open("movie.bgr24",ios::binary|ios::in);
unsigned char* buffer = new unsigned char[921600];

while(infile.read((char *) buffer,sizeof(buffer))){

cv::Mat img(Size(640,480),CV_8UC3, &buffer, 640*3);
imshow ("ch-0", img);

}

Problem:

 The imshow function in the above program is throwing the segmentation fault error.

could you help me to address this issue or I would be very helpful if you can share raw video file and OpenCV sample code to read the frames from the video file?

 

 

0 Kudos
Highlighted
Observer
Observer
138 Views
Registered: ‎04-23-2020

@idiotic_genius 

Hi

Thanks for your suggestion.

But I confused about  raw video files format.

Do It mean I can use this tool convert video to x264 format and I use openCV to read it?

 

yumei

 

 

0 Kudos