I have deepstream sdk 7.0 installed on my jetson agx orin and as per the following documentation Using a Custom Model with DeepStream — DeepStream documentation 6.4 documentation i was trying to find the example of FastRCNN which as per the above documentation should be at the following path: /opt/…

You can try to copy the data to a buffer according to the row and col of cv::Mat first. Then use the cudaMemcpy to copy the data from CPU to GPU. If there’s still a problem, it is recommended that you debug the nvinfer open source as provided earlier.

I have gone through the link you have provided and also understood the implementation. However in that it is passing landmarks while I am trying modify the input image. I have got my modified image in cv mat and then I tried to copy it in converted frame pointer. I have got the output like the follo…

You need to add the data to the pDst pointer float * pDst = (float*)buf->memory_ptr; in the CustomTensorPreparation I attached before.

as you can see in below snipped i am adding the data to pDst as you suggested but i am getting black images only. cv::Mat rgb_image; // empty Matrix cv::cvtColor(warp_mat, rgb_image, cv::COLOR_BGR2RGB); // warp_mat has my image but in BGR so performing color conversion and saving i…

UPDATE: as my model expects fp32 normalised images i modified the code as follows but still i am getting improper output. // warp_mat now has my RGB uint8 image cv::Mat float_image; warp_mat.convertTo(float_image, CV_32FC3); //converting uint8 to fp32 // Norma…

Are the images you get from cv::Mat rgb_image normal? How do you get the output image?

As i told you in above update i am storing my required image in cv::Mat warp_mat which has RGB uint8 3x128x128 format but i am converting it fp32 RGB format and normalising it by dividing with 255 in cv::Mat float_image. I am saving the output images after model inference using custom parser and i…

here is the preprocess config file i am using and preprocess.txt (2.6 KB) let me brief you again my current whole process step1: acquiring buffer from memory pool step2: using inbuf to get the input buffer surface and then copying it to intermediate temporary surface created using same parameter…

actually i have already verified image at each step like my input frame in yuv format then in rgb format then my cv modified image in rgb then in float, all of these are proper images so i the image being passed in the cudamemcpy is not an issue thus i have 2 questions: Is cudamemcpy not able to …

[image] aniketrohara: Is cudamemcpy not able to copy my image data properly? Yes, it can copy properly. [image] aniketrohara: So is it possible that issue is maybe because of this? You can try that. As you said I am saving the output images after model inference using custom parser. …

Cannot find the objectDetector_FastRCNN example

Accelerated Computing Intelligent Video Analytics DeepStream SDK

yuweiw October 9, 2024, 8:37am 36

You can try to get the image from each of your steps to see if that is normal.
Since you have verified it by using your model alone as PGIE and our nvinfer is open source. You can also get the image from each steps in the nvinfer. We have a detailed source code diagram that you can refer to.

Topic		Replies	Views
Cannot obtain Classifier raw tensor output or classifier meta DeepStream SDK tensorrt , cuda , jetson-inference , gstreamer , jetson , deepstream	20	427	February 14, 2025
How to pass custom input to non image layer of model during runtime DeepStream SDK cuda , jetson-inference , gstreamer , jetson , deepstream	13	369	December 13, 2024
Create Deepstream plugin that encapsulates process of my ONNX model - where to start? DeepStream SDK	13	1166	January 22, 2024
Too much frame drop in deepstream pipeline DeepStream SDK cuda , jetson-inference , gstreamer , deepstream	20	459	February 12, 2025
Object detection pre-trained model inference issue in deepstream DeepStream SDK tensorrt , jetson-inference , gstreamer , python	50	1275	August 9, 2024
How to transfer your own data in DeepStream DeepStream SDK opencv , gstreamer , deepstream	9	244	October 23, 2024
Custom detection ONNX model gives wrong outputs using nvinfer with DeepStream 5.1 DeepStream SDK	16	3135	September 27, 2021
Pre-processing input image with Custom object model detection DeepStream SDK	6	771	March 5, 2021
SGIE,transfer image to Device from OpenCV custom preprocessing DeepStream SDK	7	1906	March 22, 2021
How to custom preprocess in SGIE base on Deepstream 5.0? DeepStream SDK	14	3603	April 20, 2021

Cannot find the objectDetector_FastRCNN example

Related topics