I installed deepstream5.0 docker container and I ran inference with yolov3_mobilenet v2 model with help of deepstream sample apps. Here i got the bbox as output. My problem is i want to get the image crop from nvinfer stage and give it as input to zbar plugin in gstreamer. zbar plugin takes input as video streams, But i want to give the detected barcode image crop from the nvinfernce.
Can you let me know What are all the plugins we need for communication between the nvinference output and zbar plugin
Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU)
• DeepStream Version
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs)
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)
Hardware Platform - GeForce GTX 1660
Deepstream version - 5.0
TensorRT version - 184.108.40.206
NVIDIA GPU Driver Version - 460.73.01
I have installed deepstream 5.0. My work has 3 process,
- I have to detect barcodes from input image with TLT model
2)Crop the detected barcode from the image withbbox coordinates
3)Pass the cropped image into zbar plugin and read barcodes with zbar
Here I have done the first process with the help of deepstream_test3_app from /opt/nvidia/deepstream/deepstream-5.0/streams/sample_apps/deepstream-test3/deepstream_test3_app.c and i got the bounding box for the detected crops.
So Now i want to crop the detected barcodes and need to fed it as input to Zbar plugin. Here Zbar plugin takes input as video streams and this is the command line for zbar pipeline
gst-launch-1.0 -m v4l2src ! videoconvert ! zbar ! videoconvert ! autovideosink.
But My problem is I want to give the detected barcode crop to zbar pipeline. So Can you let me know What are all the plugins we need to give the nvinference output to zbar plugin.
Is it possible to implement zbar as nvinfer SGIE?
Can we send the detected crops to c++ application and process it from there?
Then how we can pass the image matrix and bbox corrdinates from meta data to c++ code to proceed further?
Is it possible to add one probe function to get gstbuffer and add your customized process in the probe function?
Yea I created dsexample and attached the output in user mata.Thank you
So you resolved your issue. right?
Yea that issue solved,But I have one more doubt. When i save the image crops after nvinfer with the code mentioned in this link How to crop the image and save - #9 by iyaqoobi, i can’t able to save the original crops without padding because its scale the crops to the default image width and height mentioned in dsexample. But i want to save crops without padding, So How to save the detected object crops without padding in dsexample?
dsexample is one sample code. you can change it. Maybe you don’t need keep aspect ratio.
Thank you @kesong ,I have done that. and one more issue, the current pipeline (deepstream_test3_app from /opt/nvidia/deepstream/deepstream-5.0/streams/sample_apps/deepstream-test3/deepstream_test3_app.c) only supports low resolution (720p) video. and the pipeline is not processing when i gave high resolution (12MP) video as input. Is there any solution to give high resolution video?
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.