Shulou(Shulou.com)06/01 Report--

This article mainly explains "how to use OpenCV". The content in the article is simple and clear, and it is easy to learn and understand. Please follow the editor's train of thought to study and learn how to use OpenCV.

Have you ever thought of letting the Mona Lisa follow your expression and have a cordial exchange?

The first-order motion model (First Order Motion Model) of Aliaksandr can be implemented, but the implementation process is very complex and tedious.

The first-order motion model is powerful and can generate special effects on images and videos on untrained data sets, but at the cost of tedious installation and configuration.

Can you make it a little easier, a little simpler?

An Indian programmer, Anand Pavara (Anand Pawara), designed real-time animation effects based on OpenCV.

After all, OpenCV is a well-known cross-platform visual library and a de facto standard library in the field of computer vision.

Face recognition, motion recognition and motion tracking involved in the project are all natively supported in the OpenCV library.

Installation process

1. Install the dependency module

Install the dependency module:

Pip install-r requirements.txt

Install pytorch 1.0.0:

Pip install torch===1.0.0 torchvision===0.2.1-f https://download.pytorch.org/whl/cu100/torch_stable.html

2. Download the configuration file (if it cannot be downloaded, the file link is at the end of the article)

Gdown-id 1wCzJP1XJNB04vEORZvPjNz6drkXm5AUK

3. Run the program

Run the file:

Python image_animation.py-I path_to_input_file-c path_to_checkpoint

Real-time special effects for cameras:

Python.\ image_animation.py-I.\ Inputs\ Monalisa.png-c.\ checkpoints\ vox-cpk.pth.tar Run application from video file: python image_animation.py-I path_to_input_file-c path_to_checkpoint-v path_to_video_file

Special effects generation for existing videos:

Python.\ image_animation.py-I.\ Inputs\ Monalisa.png-c.\ checkpoints\ vox-cpk.pth.tar-v.\ video_input\ test1.mp4

If you want to get started, just adjust the relevant configuration file.

The model is divided into two usage modes, one is the more conventional method of importing video, and the other is to generate video special effects in real time.

Note, however, that you must use pytorch version 1.0.0, because later versions have problems with first-order models that the back end depends on.

According to the author's follow-up plan, the client program will be launched and the falsetto (deepfake voice) function will be added.

Optimization based on OpenCV

Anand's work is to simplify the existing first-order motion model (First Order Motion Model) and use OpenCV to generate special effects for video.

The back-end support of the project is done by the OpenCV library, avoiding the use of complex models and lowering the barriers to use.

According to the author's point of view, the use of this model only needs to train a class of data sets, and then it can be applied to all its things, that is, it has a better generalization ability.

Different from the test results of multiple data sets of the original first-order model, Anand's real-time dynamic special effects model is only tested on face data sets, and other data sets continue to be added later.

The characteristic of this model is easy to use, easy to configure, basically can be used out of the box, even training your own data set will be relatively simple.

Because it is based on the optimal allocation of existing resources, it is easy to operate and powerful.

Of course, simplicity can also bring problems, such as the current data set is relatively single, mainly aimed at the face data set.

Thank you for your reading, the above is the content of "how to use OpenCV", after the study of this article, I believe you have a deeper understanding of how to use OpenCV, and the specific use needs to be verified in practice. Here is, the editor will push for you more related knowledge points of the article, welcome to follow!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.