Hello trend,
This is Satya Mallick from LearnOpenCV.com. Today's blog post is the definite guide to OpenCV DNN module but before we go into the details, I have a quick announcement.
Until last week, anybody who wanted to enroll in Official AI Courses by OpenCV had to join a waitlist. Now, you can purchase instantly with no discount or join the waitlist for a discount. With that out of the way, let's learn more today's post on OpenCV DNN module.
We used to think of OpenCV as the library for classical computer vision algorithms. That changed in 2016 with the addition of OpenCV DNN module which enabled inference using Deep Neural Networks (hence, the name DNN).
Not only is this module super simple to install it also offers support for almost all frameworks like PyTorch, TensorFlow and the generic format - ONNX.
We will first compare the performance of the OpenCV DNN module with PyTorch and TensorFlow for image classification and object detection tasks. We will also see how to use pre-trained models for real-time inference on Intel CPU using the OpenCV DNN module. Without further ado, lets dive into the post
https://learnopencv.com/deep-learning-with-opencvs-dnn-module-a-definitive-guide/
As always, we are sharing code in Python and C++ at the link below
https://github.com/spmallick/learnopencv/tree/master/Deep-Learning-with-OpenCV-DNN-Module
OpenCV DNN Module on NVIDIA GPU
From the very beginning OpenCV DNN Module was designed to be blazingly fast on CPUs (see data in the post), and users had the option to choose different backends like OpenVINO. GPU support for OpenCV DNN module was sorely lacking. So in 2019, I reached out to Davis King (author of Dlib) for help. He generously agreed to mentor Yashas Samaga who added GPU support to OpenCV as part of Google Summer of Code 2019.
If you want to try the DNN module with NVIDIA GPU acceleration, the following posts will be very useful
| OpenCV DNN Module with Nvidia GPU on Windows | Code |
| How to use OpenCV DNN Module with NVIDIA GPUs | Code |
GRAYSCALE VS COLOR CAMERA
As part of OpenCV Weekly Webinar, we are bringing world class practitioners to discuss practical problems and solutions related to computer vision and artificial intelligence.
In tomorrow's webinar, we will discuss whether to choose a color or a grayscale camera for your computer vision project. In the words of Brandon Minor (CEO, Tangram Vision)
This doesn't appear to be an urgent problem on its surface.
Both give visible data, right? Not quite.
Color and grayscale each have their own unique characteristics and applications. These qualities should be considered when applying computer vision to any application. To explore these concepts fully, we must first discuss how grayscale and color cameras are constructed from the CMOS sensor on up; how each pass and process information; and what applications can benefit from each camera modality given these constraints.
Fascinating! When you talk to an expert, you often realize how much there is to learn, and how much these little details count.
Topic : In Living Color (Or Not), Grayscale vs. Color Cameras
Time : 9 AM Pacific Time, April 15, 2021
Hosts : Brandon Minor (CEO, Tangram Vision), Satya Mallick (CEO, OpenCV.org), and Phil Nelson (Content Manager, OpenCV.org)
Free registration : Zoom Registration Link
You can check out last week's episode at the link below
I hope to see you tomorrow!
Satya
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.