The Easiest Way To Do Object Detection in C#

[@ProgrammingWithChris]

Thanks for watching! If you have anything else in the OpenCV library that you'd like to see done in C#, please let me know! 🕘Timestamps: 0:00 - Video Start 00:51 - The Haar Cascade Classifier Code 05:39 - Haar Cascade Classifier Demo 06:49 - The YOLO Object Detection Algorithm in C# 27:31 - YOLO Object Detection Demo

[@ozzietradie6514]

objection detection is pretty powerful.When combined with a box around the target, (simple logic "like when box A meets box B" do something) for example object A is a beer glass object B is the table when A rectangle intersects B rectangle order another beer.

[@s1mplename]

why tho. whats ur point

[@RaabStephan]

This is really cool. Would be cool to use OpenCV to do apriltag detection and tracking. You could even have the apriltags read different colors (a red code, a blue code, a green code) and then based off of the template you could draw the correct color around the object as you're tracking it.

[@Ny_babs]

how are you setting your camera to be in source for the analysis? I think I missed that step somewhere. Great vid!

[@jajnelumstransport2624]

how would I get the name from of the detected object and store it in a value. Thanks.

[@AJ93-lx6on]

can you also do tutorial on manual train a model and then add that to project

[@atincuzun]

hi chris, great video but i still have some questions in emgucv. i am having some troubles when i try to train custom model for detection. emgucv gives me error "number of input channels should be multiple of 3 but got 256" etc. i still don't get how to set the inputs for cnn individually(emgucv mat and arrays are really confusing since you dont know how they are laid in the memory, documentation does not help either). for example, how can i pass an image to the cnn if the layer inputs are different than shown here?

[@ProgrammingWithChris]

Yeah, that's an issue I ran into as well, I believe. At least, my problem was the net.Forward was expecting a very specific dimension on the input. If your issue is similar to the one I'm thinking of, I'd recommend playing with the .Resize of the frame, even just manually putting the pixel values in the System.Drawing.Size() parameter to it so that you guarantee that it's a multiple of 3 (just have to do some math there). If that doesn't help, let me know and maybe point me to some code so I can take a look.

[@lot10games]

I get this error : Emgu.CV.Util.CvException: 'OpenCV: Transpose the weights (except for convolutional) is not implemented'

[@lot10games]

yolov3.weights is 2Kb LOL. It's corrupted or something . I downloaded from other source and it works.

[@narf0339]

which language get you further or can do more for object tracking ? c# or python ?

[@minhthiennguyen2740]

how to capture image and detect ? I don't want detect when wedcam run live stream?

[@كنوزإمامالدعاة]

I am planning to build a software that can help blinds to detect objects and faces, to what extent this can be helpful? or I should think of different library/language/framework?

[@peri.platypus]

Is it possible in plate number? Using c#? What is best to use haar classifier or yolo? Thanks

[@JustBrandonLim]

Hi, thanks for this great tutorial. I am currently trying out to run inference from an ONNX model, and I seem to have problems finding a way to unwrap the VectorOfMat that is returned after running a forward pass. Is there any documentations on unwrapping?

[@aidanjones2885]

Hi Chris, I tried to run the source code for the YOLO example and I added those 4 files to my project file but I get the error: An unhandled exception of type 'Emgu.CV.Util.CvException' occurred in Emgu.CV.Platform.NetStandard.dll Additional information: OpenCV: Failed to parse NetParameter file: ./detection/yolov3.cfg. This happens by the first line var net = Emgu.CV.Dnn...etc. Do you possibly know how to solve this problem? Am I putting those files in the wrong place?

[@ProgrammingWithChris]

Are the files in a folder inside of your project called /detection/ ? if not, you'll just have to adjust to point it to the right location.

[@digitarts6085]

I had same problem, but I was able to solve it by installing Emgu.cv.runtime.windows to project.

[@PhạmNguyễnHoàngAnh-anhpnh]

Helo, how to Object Tracking on C#?

[@lanaguilfoyle8072]

Hi Chris, really helpful and great tutorial! I have errors on the first line with var net = Emgu.Cv.Dnn.. when I try to run the program: System.TypeInitializationException: 'The type initializer for 'Emgu.CV.Dnn.DnnInvoke' threw an exception.' And: DllNotFoundException: Unable to load DLL 'cvextern' or one of its dependencies: The specified module could not be found. Is there any chance you can help?

[@abbeykirkman397]

did you fine the solution to this?

[@digitarts6085]

@@abbeykirkman397 I had same problem, but I was able to solve it by installing Emgu.cv.runtime.windows to project.

[@PixeI]

Wauww nice content! Can you make a video showing YOLOv6 or 7 working with C#?

[@peri.platypus]

You make it look easy😂

[@GR4MPI]

Hi, do you think that it could be used to capture all the chess pieces and their location, and then return their positions in 2d array representation of chess board ?

[@kerverse]

Any updates? Ive been working on a similar project and dont have any leads at the moment

[@s1mplename]

@@GR4MPI well you could make a sort of detection using a color based analysis. though you would need to really work on a project and create a tech from scratch

[@GR4MPI]

@@s1mplename yes.. i did my research and decidedd not to go this way :)

[@s1mplename]

@@GR4MPI i was tasked with a project about a new type of age detection. Im still pondering about how to make it. Its tooo hard 😭

[@GR4MPI]

@@s1mplename the harder the fight is, the greater is the victory. Stay strong bro 💪

[@azhanisyahputra9867]

This is educational 😍

[@qpumuJIbek]

Can u do "The Easiest " code in 10-15lines to demonstrange to how EASY start. no those things. omg. BE SIMPLE DUDE! there mose ppls who cant/dont want read documentation.