Deepstream-imagedata-multistream py saving corrupted frames into the folder

[B_R_191]

• Hardware Platform:Jetson(orin)
• DeepStream Version: 7
• JetPack Version :6.0
• TensorRT Version:8.6.2.3
• NVIDIA GPU Driver Version
• Issue Type:question
i am testing deepstream-imagedata-multistream py provided in /apps/deepstream-imagedata-multistream/ . i have tested it on custom model i am able to see inferance video running but when i see my folder where extracted frames are stored , i am getting below frame

[yuweiw]

What video source do you use? Is there the problem with just one picture, or all of the pictures.

[B_R_191]

i am using h264 video for inferance and getting all frames like above one

[yuweiw]

Could you attach your h264 file and use our demo file to have a try?

/opt/nvidia/deepstream/deepstream/samples/streams/sample_720p.h264

[B_R_191]

@yuweiw

samples

i have tried using video you suggested still i am getting same type of frames

Note:[Earlier, i had commented draw_bounding_boxes function hence in first frame there is no bounding box]

[yuweiw]

Could you attach your specific model of the board, like AGX Orin , Orin NX or Orin Nano?

[B_R_191]

I am using Orin Nano.

[yuweiw]

I use Orin Nano to run our demo code deepstream-imagedata-multistream. The image is normal.
Have you modify our source code?

[B_R_191]

I am using custom model and i have make changes in the tiler_sink_pad_buffer_probe function . I also tried with default code but it is giving me segmentation fault or free(): invalid next size (fast)
Aborted (core dumped)
error. I am attaching modified code file
deepstream_imagedata-multistream.txt (20.9 KB)

[yuweiw]

Let’s narrow down that first.

1. Are there any problems with both our demo code and model?
2. Are there any problems with our demo code and your model?

[B_R_191]

1.I am getting problem with your demo code and your model .Problem is, after processing some frames i am getting segmentation fault or free(): invalid next size (fast) error.
I am attaching two txt file which contain terminal output just take look into that
output.txt (4.5 KB)
output2.txt (4.5 KB)

2 Also when i am using my model with your code also i am getting same error mentioned in txt file

[yuweiw]

OK. Let’s use our demo and model to analyze the crash problem first. Because the same platform on my side runs that normally, it may be the problem of your environment. Could you use gdb tool to do a preliminary analysis?

$gdb --args <your_command>
$r
after the crash
$bt

[B_R_191]

Here is the screenshot after running code with gdb tool

[yuweiw]

Could you try it with our source code without any changes? You can try to remove some of your extra definitions, like pgie_class_str, obj_counter, etc… .

[yingliu]

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks