100+ FPS Object Detection: How to Deploy Neural Networks with OpenCV DNN and GPU in C++

You will also get access to all the technical courses inside the program, also the ones I plan to make in the future! Check out the technical courses below 👇

_____________________________________________________________

In this video 📝 we are going to learn How To Deploy Neural Networks with OpenCV DNN and GPU in C++. We will go over each step in deploying trained neural networks for inference with OpenCV. We will see how to use the methods to load in neural networks with OpenCV and deploy them. At the end of the video, we will see an example with object detection, where we can detect a lot of different objects from the Coco dataset with 100 FPS+ on a GPU with OpenCV.

If you enjoyed this video, be sure to press the 👍 button so that I know what content you guys like to see.

_____________________________________________________________

_____________________________________________________________

_____________________________________________________________

📞 Connect with Me:

_____________________________________________________________

🎮 My Gear (Affiliate links):
🖥️ Desktop PC:

_____________________________________________________________

Tags:
#OpenCVdnn #OpenCV #DNN #ObjectDetection #ComputerVision #DeepLearning

[NicolaiAI]

Join My AI Career Program
Enroll in My School and Technical Courses

[vasilimacharadze]

any link for downloading model files (e.g. frozen_inference_graph.pb..)?

[CosmJJ]

Amazing channel and content, man! Thank you soo much!

[ethanshay]

I just found this channel, keep up the content! I subscribed lol

[pietrarualfred]

Hello, where can I get those binaries in order to reproduce the tutorial on my own device?

[martuscellifaria]

Hi Nico, I was trying to reproduce it by myself, but I became stuck after running into this specific error.

global (935) open OpenCV | GStreamer warning: Cannot query video position: status=0, value=-1, duration=-1
terminate called after throwing an instance of 'cv::Exception'
what(): OpenCV(4.2.0) error: (-2:Unspecified error) Can't create layer of type "AddV2" in function 'getLayerInstance'

I really don't know what has passed, since it never occurred on Python. It is basically saying that I can't use my webcam for streaming, because of GStreamer, am I right? Any clues on how to fix it? I am running it in Ubuntu 20.04, if it helps.

[rukmanivinothkumar]

Hi, i m getting error :(215: Assertion failed) !empty() in function Could u give me a solution for resolving this error

[brandonsamuelcruzsilva]

Hi! I haven't watch your previous videos so I'm not familiar with how you deploy your GPU. I just wanted to know, are you are using the GPU locally?

[luchex]

Excellent video! something that i noted is that c++ program is slower than python program, thats correct?

[yashas]

What GPU did you benchmark on? What version of cuDNN did you use?

OpenCV CUDA backend has performance problems with depthwise convolutions. MobileNetSSD makes heavy use of depthwise convolutions and hence performs really bad on OpenCV CUDA backend. I get 100+ FPS on Mask RCNN (1024x1024 image), 120+ FPS for YOLOv4 (608x608), and ~2000FPS for YOLOv4 Tiny on 2080 Ti. MobileNet must in principle performance vastly better than the aforementioned models but it barely manages to get past 100 on 2080 Ti!

There is also a FP16 target `DNN_TARGET_CUDA_FP16` that can further boost the performance if your GPU supports it.

[OUKSEE]

Thanks for the video! I am getting an error and have not been able to resolve the issue from other threads online.
Any ideas for "OpenCV(4.5.3-dev) Error: Unspecified error (Const input blob for weights not found)" error?

[peterSobieraj]

Can I download that network somewhere ?

[SKDong-czno]

I can't find the pb model in your github

[dasich]

so, describing opencv example WITHOUT the instruction(to make that 2 lines working) on how to compile opencv with cuda dnn support enabled....

[shaunjohnsen]

HI! I am currently working on a security surveillance system as a side project. When I try to load the file, then I get an exception stating that it can't read the file, any suggestions? I am sure that my file path is correct.

[kevinrafi]

It doesn't work on me. this error occurs when i run the code:

[ERROR:0] global (2562) parseNode DNN/TF: Can't parse layer for Exception: OpenCV(4.5.2) error: (-2:Unspecified error) Const input blob for weights not found in function 'getConstBlob'

terminate called after throwing an instance of 'cv::Exception'
what(): OpenCV(4.5.2) error: (-2:Unspecified error) Const input blob for weights not found in function 'getConstBlob'

Aborted (core dumped)