The latest deep learning technology, just follow along!
An Easy Guide to 2D Pose Estimation 🤖
In my videos and images
How to estimate poses in one go! 📖

2D pose estimation is a popular topic, ranked among the top five keywords in computer vision, and is fundamental to most technologies. In this lecture, we'll learn about pose estimation models using the latest LitePose and DCPose models, and how to train these models using real-world videos.
By the way, what is 2D pose estimation? 🧐

2D Human Pose Estimation (HPE) is the task of estimating the 2D coordinates of human joints from images or videos. This technology is used as a foundational technology in a variety of fields. It is a crucial research area, with applications not only as a backbone model for 3D pose estimation, but also in the animation industry, virtual reality and augmented reality (VR/AR), and even in analyzing the movements of sports players.
Pose estimation models are a fundamental field that has been studied for a long time and is still being studied in various ways, so you can learn various deep learning techniques more easily after taking the course.
Implement 2D pose estimation technology yourself!
This course requires basic knowledge of Python and deep learning . It is designed for beginners or those interested in pose estimation projects. It aims to help you understand and practice pose estimation models . After taking this course, you'll be able to implement 2D pose estimation techniques on your own computer.
The important thing is that I will teach you step by step, from installing Ubuntu to running the code !
- 💡 Leverage the latest 2D pose estimation models for videos of interest, such as golf and yoga.
- 💡 How to use pose estimation in deep learning projects like object recognition
I recommend this to these people 🚩
✅ People who don't know 2D pose estimation but want to study the latest papers and try setting up and developing the environment
✅ People who are tired of Object Detection projects and want to start a new project
✅ People who want to quickly apply a pose estimation model to projects such as graduation projects or competitions
✅ People who don't want to go through trial and error when starting pose estimation research
Preview the lecture 📺
Introduction to 2D Pose Estimation → LitePose Paper Theory and Practice → DCPose Paper Theory and Practice → Custom Dataset Practice
You can start with the overall flow of 2D pose estimation research and proceed with a custom learning process that you can apply directly to your own data .
We'll highlight the core theories discussed in papers presented at the top conference, CVPR 2021-2022.
Even if you're new to 2D pose estimation, it's OK! The model training process is easy to follow, with easy-to-follow steps.
The person who created this lecture, Roasted Chestnut🌰
Based on my knowledge of computer science and artificial intelligence, experience with numerous deep learning/machine learning projects, and graduate school research experience, I will provide you with essential information.
- Master's degree in Artificial Intelligence from Korea University
- Numerous SCI(E) papers and presentations at international conferences
- Working as an AI researcher at a major corporation (Computer Vision Research)
I completed the combined bachelor's and master's degree program in the Department of Artificial Intelligence at Korea University in just three semesters , and published a paper on 2D Video Human Pose Estimation at WACV2023 (Winter Conference on Applications of Computer Vision ) , the world's 9th largest conference in the field of artificial intelligence and computer vision , and was selected for an oral presentation.
Q&A 💬
Q. Can I take this course without any prior knowledge of pose estimation?
Yes! Even if you don't have a basic foundation, you don't have to worry because we'll walk you through how to run the code from start to finish. If you're simply interested in pose estimation, you can take this course. (However, you should have a basic understanding of deep learning and Python syntax !)
Q. What can I do if I learn pose estimation?
Pose estimation remains an active research area and is utilized across a wide range of industries. It currently underpins a variety of technologies, including drowsy driving detection, pose estimation in AR and VR environments, and coaching in the sports industry. The possibilities are endless!
💾 Please check before taking the class!
- The course will be conducted in an Ubuntu environment equipped with GPU and CUDA development environments, as models are trained based on PyTorch and actual papers.
- This lecture was recorded using Linux and Ubuntu 20.04 . We recommend using a compatible environment.
- To practice implementing the model, you will need a PC with a graphics card .
- All source code used in the lectures is provided to students as learning materials.