• Hardware Platform: GPU • DeepStream Version: 5.0.0 • TensorRT Version: 7.0.0.11 • NVIDIA GPU Driver Version (valid for GPU only): 460.32.03 Hi, I have a pipeline including ... -> pgie -> queue -> sgie -> ..., where pgie is a detector and the cropped detections are fed into sgie. However the ou…

The problem in my case was that PGIE outputted a gray scale image with values ranging from 0 to 255. However my model expected a float input between 0 and 1. So the solution was to scale the pixels on input using net-scale-factor=0.003921568859368563.

did you use samples to reduced this problem? this happened in several topics related problem with performance of the secondary engine output. U can see when search on forum, but have not hear any responded for that yet. [+1 concern]

[image] trild-vietnam: did you use samples to reduced this problem? Can you elaborate that? I don’t understand it. What I did so far is, I deconstructed my problem to not take a video as input but an image and found an image which has just one detection in PGIE so I would get a minimal setup …

[1] use sample convert buffer to save crop image from buffer by opencv in function get_converted_buffer of gst-nvinfer. [2] I found this thread. [image] Reduced detection accuracy of the secondary model DeepStream SDK • Hardware Platform (Jetson / GPU) T4 GPU …

I wanted to have the crop which is being fed into SGIE. So what I did is the following: In gstnvinfer.cpp’s gst_nvinfer_process_objects method, after get_converted_buffer is called, as I wanted to make use of the actual used transformation parameters, I added the following code: //create temporar…

yes, I did it to debug too. Then how did it work?

I took this code to create the cropped PGIE output and save it as file. It looks like I expected it. Further on I removed the SGIE from my pipeline and configured the PGIE with the same parameters I used for the SGIE previously and fed the cropped image in. However, the output is the same. No idea w…

can check more if you post your config-file both pgie end sgie here

Well, my PGIE config for the cropped input looks like this: [property] gpu-id=0 net-scale-factor=1 gie-unique-id=1 onnx-file=model.onnx batch-size=200 process-mode=1 ## 0=FP32, 1=INT8, 2=FP16 mode network-mode=0 model-color-format=2 network-type=100 output-tensor-meta=1 output-blob-names=import/0…

I did that but the created libnvds_infer is not even linked to my app. So the code I edited there is not reached at all.

Debug nvinfer input

Accelerated Computing Intelligent Video Analytics DeepStream SDK

bcao March 5, 2021, 8:07am 11

Hey, you can refer DeepStream SDK FAQ - #9 by mchi to dump the input

Topic		Replies	Views
How images from custom post-processing can be used as input for secondary GIE？ DeepStream SDK	6	453	September 2, 2022
How to debug gstnvinfer with custom model？ DeepStream SDK tensorrt	23	2522	June 11, 2021
How to visualize input patches of sgie DeepStream SDK deepstream	1	398	December 1, 2023
Try to get the result of SGIE result, failed DeepStream SDK deepstream61	8	758	July 24, 2022
NvBufSurfTransform failed with error -2 while converting buffer DeepStream SDK	1	1835	October 9, 2020
Can only see SGIE inferences inside the PGIE inferences. Not sure if SGIE is acting on full frame DeepStream SDK	11	699	September 26, 2023
Secondary GIE (ONNX gaze model) never attaches tensor-meta (buffer always NULL) DeepStream SDK deepstream	6	234	May 10, 2025
DSExample mutiple inferencing after PGIE DeepStream SDK	1	648	August 5, 2020
DeepStream nvinfer input tensor contains incorrect image DeepStream SDK jetson-inference , gstreamer	13	1628	August 8, 2022
How to custom preprocess in SGIE base on Deepstream 5.0? DeepStream SDK	14	3603	April 20, 2021

Debug nvinfer input

Related topics