System Information

OpenCV version: 4.10.0 & 5.0.0
File: yolo_detection.cpp (4.x version) / object_detection.cpp (5.x version)

Detailed description

In the OpenCV 4.10.0, within the yolo_detection.cpp file at line 224, the Image2BlobParams object is instantiated as follows:

Scalar scale = parser.get<float>("scale");

It should be changed to use the Scalar::all() function like this:

Scalar scale = Scalar::all(parser.get<float>("scale"));

Similarly, in the OpenCV 5.0.0-pre version, within the object_detection.cpp file at lines 402 and 606, there are comparable implementations where the global variable scale is of type float. In these cases, the Scalar::all() function should also be utilized to apply the scaling factor correctly across all channels.

At line 401, I think the below is correct:

        Image2BlobParams imgParams(
            Scalar::all(scale),
            size,
            meanv,
            swapRB,
            CV_32F,
            DNN_LAYOUT_NCHW,
            paddingMode,
            paddingValue);

Also at line 605:

            Image2BlobParams paramNet;
                paramNet.scalefactor = Scalar::all(scale);
                paramNet.size = Size(inpWidth, inpHeight);
                paramNet.mean = meanv;
                paramNet.swapRB = swapRB;
                paramNet.paddingmode = paddingMode;

Steps to reproduce

Even though using the current code doesn't produce an error message, reviewing the implementation of the Image2BlobParams class shows that it's correct to use the Scalar::all() function when a float argument is passed.

Issue submission checklist

• I report the issue, it's not a question
• I checked the problem with documentation, FAQ, open issues, forum.opencv.org, Stack Overflow, etc and have not found any solution
• I updated to the latest OpenCV version and the issue is still there
• There is reproducer code and related data files (videos, images, onnx, etc)