Human Pose Estimation


  • Machine Learning

  • Image Processing


Mentors :

  • Om Godage

  • Shubham Hazra

Mentees :

  • 7


Pose estimation is a fascinating field that deals with the accurate and reliable detection of human body movements and positions. It is widely used in various applications such as robotics, gaming, and virtual reality. The problem involves analyzing a given video or image to determine the position and orientation of the human body in real-time.
To implement this, one can use computer vision techniques such as deep learning and convolutional neural networks. By training these networks on large datasets of human poses, they can accurately identify the positions and movements of the human body in various scenarios. The result is a highly accurate and efficient system that can be integrated into a wide range of applications, from sports coaching to medical rehabilitation.
With the increasing demand for accurate and real-time human pose estimation, this project offers a unique opportunity to explore the latest techniques and tools in computer vision and machine learning. The implementation of a robust and efficient pose estimation system can have a significant impact on various industries, from gaming to healthcare, and can pave the way for exciting future developments in the field.
Here is a link for some general information: https://www.v7labs.com/blog/human-pose-estimation-guide
We are going to select mentees based on the assignment given below. Give it your best shot!: https://github.com/0-JackFrost-0/Human-Pose-Estimation-SoC-2023 Pre-requistes:

  • Python
  • Basic Mathematics

Tentative Timeline :

Week Number Tasks to be Completed
Week 1-3 Basics of Regression & Classical ML, Intro to Deep learning & frameworks (Tensorflow, PyTorch), Image Processing using OpenCV & classical methods
Week 4-6 Dive into CNNs & transfer learning, Intro to transformer and attention models & pose estimation, Getting Started with Pose Estimation
Week 7-9 Implementing Pose Estimation using State-Of-The-Art Models, Finishing up with the motion detection pipeline and applying on new data, Finishing up with documentation and submission