- 1 Introduction
- 2 Computer Vision GitHub Repositories
- 2.1 3D Face Reconstruction using CNN ( ★ – 4.1k | ⑂ – 682 )
- 2.2 Real-time Multi-Person Pose Estimation and Tracking System ( ★ – 4k | ⑂ – 1.1k)
- 2.3 Automatic Colorization of Photos using Deep Neural Networks (★ – 2.3k | ⑂ – 560 )
- 2.4 Editing natural photos using Generative Neural Networks ( ★ – 1.9k | ⑂ – 186 )
- 2.5 Convolution Recurrent Neural Network for Image Recognition ( ★ – 1.7k | ⑂ – 486 )
- 2.6 Image Deblurring using Generative Adversarial Networks ( ★ – 7.8k | ⑂ – 1.8k )
- 2.7 Painting AI – Deep Reinforcement Learning model that produces paintings using strokes ( ★ – 1.7k | ⑂ – 227 )
- 2.8 Lip Reading – Cross Audio-Visual Recognition using 3D Architectures ( ★ – 1.4k | ⑂ – 246 )
- 2.9 Quickdraw – Interactive drawing identification tool ( ★ – 677 | ⑂ – 107 )
- 2.10 Image Animation using First Order Motion Model ( ★ – 3.9k | ⑂ – 780 )
- 2.11 Fashion MNIST ( ★ – 7.8k | ⑂ – 1.8k )
- 2.12 Cool Computer Vision Projects ( ★ – 37 | ⑂ – 45 )
- 2.13 Intermediate Level Computer Vision Projects ( ★ – 13 | ⑂ – 9 )
- 3 Conclusion
Though GitHub is a version controlling and open source code management platform, it has become popular among computer science geeks to showcase their skills to the outside world by putting their projects and assignments on GitHub. You can find many amazing GitHub repositories with projects on almost any computer science technology, uploaded by people or teams. In this post, we are listing down some awesome computer vision GitHub repositories that should inspire you to build your own project like this.
We are also listing down the stars (★) and the number of forks (⑂) these GitHub repositories have got (at the time of writing this) to give you an idea of their popularity.
Computer Vision GitHub Repositories
13D Face Reconstruction using CNN ( ★ – 4.1k | ⑂ – 682 )
This GitHub repository has a project where Convolutional Neural Networks are used to reconstruct 3D Face Models using 2D images. It’s a comprehensive repository where we have the option to work with this model using different languages like MATLAB, Python, etc. To make it more interesting, we can even use our own images or our own examples and test them over this model.
2Real-time Multi-Person Pose Estimation and Tracking System ( ★ – 4k | ⑂ – 1.1k)
This real-time multi-person pose estimation tracking system is called AlphaPose. This system is basically a way to map the movement of individuals in real-time. Apart from this, it can also estimate the pose individuals will make. Again this repository can help in understanding the deeper insights about how such system works. This can also be a starting point for building such software that leverages this kind of pose estimation and tracking ability.
3Automatic Colorization of Photos using Deep Neural Networks (★ – 2.3k | ⑂ – 560 )
Another interesting computer vision project is this colorization of black and white photos using deep neural networks. This computer vision GitHub repository contains python code in the Jupyter notebook, making it easy to understand. Along with this, there is an abundant dataset of images for training and testing of the model built for this task.
4Editing natural photos using Generative Neural Networks ( ★ – 1.9k | ⑂ – 186 )
This repository is hosting the code for the research paper “Neural Photo Editing with Introspective Adversarial Networks“. This project comprises a simple interface where we can edit natural photos using generative neural networks. The current version is compatible with python 2.7 version, there are still some inconsistencies with the latest versions of python.
5Convolution Recurrent Neural Network for Image Recognition ( ★ – 1.7k | ⑂ – 486 )
This is a very interesting GitHub repository where you can build an image recognition system using a convolution recurrent neural network. This project is also useful in building scene text recognition and optical character recognition. The repository contains dataset for training and testing purposes, along with this there are demo examples as well.
6Image Deblurring using Generative Adversarial Networks ( ★ – 7.8k | ⑂ – 1.8k )
A lot of times we are annoyed due to blurry images, this GitHub repository has a solution for this situation. A PyTorch implementation of the paper titled “DeBlur GAN” basically takes a blurred image as input and produces a sharp image of the input using Generative Adversarial Networks. Again the repository has the complete source code and different kinds of datasets that should assist you for better understanding and appropriate testing of the models built.
7Painting AI – Deep Reinforcement Learning model that produces paintings using strokes ( ★ – 1.7k | ⑂ – 227 )
The Painting AI GitHub repository contains a deep reinforcement learning-based model that teaches machines to paint human-painted pictures by using a fewer number of strokes. Since it is based on reinforcement learning, the project doesn’t require data for training purposes. The agents learn on their own for painting like humans. I would highly recommend you to check out the repository and try out your hands on it.
8Lip Reading – Cross Audio-Visual Recognition using 3D Architectures ( ★ – 1.4k | ⑂ – 246 )
Lip Reading is a computer vision project that looks to solve the problem encountered in audio and visual streams. This project uses audio-visual recognition for mapping the audio with the video. All of this is achieved using 3D Convolutional Neural Network Architecture for the mapping operation. This repository can surely help in building models that can fight against fake videos and other such malpractices.
9Quickdraw – Interactive drawing identification tool ( ★ – 677 | ⑂ – 107 )
Quickdraw is a computer vision project that can be used to identify a set of objects drawn using a pen (similar object). The drawing is identified through the webcam of pc and then the model tries to predict the object from the list of objects it is trained to identify. Quickdraw is basically an online game developed by Google. Another version of this project is to identify projects drawn over a canvas.
10Image Animation using First Order Motion Model ( ★ – 3.9k | ⑂ – 780 )
This is a very amazing computer vision GitHub project, here we can use our own face as a mimic to animate faces from a video or image. The model takes a driving video and maps its motion over static images to make the movement appear realistic. This same concept is applicable to the fashion dataset as well.
11Fashion MNIST ( ★ – 7.8k | ⑂ – 1.8k )
This GitHub repository consists of images for different kinds of clothes worn by people. The repository has a training set of 60,000 images and a test set of 10,000 images. Each of the images is a 28×28 grayscale image. It contains models that are built using the dataset available. Generally, this repository also helps to validate your own Machine Learning algorithms by using it on the dataset. This is a beginner-friendly dataset hence they can benefit immensely from this repository to get a feel of computer vision projects.
12Cool Computer Vision Projects ( ★ – 37 | ⑂ – 45 )
This repository hosts many interesting computer vision projects like Face Recognition, Digit Recognition, Facial Expression Detection, Object Detection, Object Tracking, etc. Through this repository, you can learn about some really cool computer vision stuff. You can take inspiration to build these projects on your own or add extend their functionalities. This will really help to learn a lot and add weightage to your profile.
13Intermediate Level Computer Vision Projects ( ★ – 13 | ⑂ – 9 )
This is yet another useful GitHub repository that has multiple computer vision projects like Hand Gesture Recognition, Face Recognition, Content-Based Image Retrieval, etc. These are good projects for the intermediate level and will help to improve your computer vision concepts.
In this article, we saw many GitHub repositories for computer vision projects to inspire you. We hope it will help you to create your own computer vision projects to amaze others and enhance your learning and resume.