How to use blobFromImages in c++

asked 2019-10-10 07:17:24 -0500

Shay Weissman gravatar image

updated 2019-10-10 07:20:24 -0500

berak gravatar image
System information (version)
  • OpenCV => :3.4.5
  • Operating System / Platform => windows 7 windows 10:
  • Compiler => :microsoft vs2019:
  • C++
    Detailed description

I am using dnn module very successfuly with yolo v3 and ssd mobilenet, with single image process - blobFromImage I want to process few images in parallel using blobFromImages. I wrote: (yolov3 net)

frame1 = imread("img1.jpg");
frame2 = imread("img2.jpg");
std::vector<cv::Mat> inputs;
inputs.push_back(frame1);
inputs.push_back(frame2);
blobFromImages(inputs, blob, 1 / 255.F, inpSize, mean, true, false);
std::vector<Mat> outs;
net.forward(outs, getOutputsNames(net));
postprocess(frame, outs, net);

When I am loading only the one image it works fine. When I am lodaing 2 images the outs matrices are empty. Waht am I doing wrong?

Steps to reproduce

edit retag flag offensive close merge delete

Comments

I have a 2 images sample program with tensorflow model that works just fine. I try to migrate it using my darknet model. It seem like the output matrix is different when using one image versus using 2 images. The output matrix when entering one image is making sense 2028 x 85 The output matrix when entering 2 images is -1 x -1. Is the output format in the 2 cases different? When using tensorflow model the output matrix size is -1x-1 but the way the output is taken is different from when darknet model is take out.

Shay Weissman gravatar imageShay Weissman ( 2019-10-13 01:34:51 -0500 )edit

I think that opencv with darknet model is not working with multiple images !!!! Can someone confirm it please.

Shay Weissman gravatar imageShay Weissman ( 2019-10-15 01:30:40 -0500 )edit