Project Overview

This is the GitHub repository for the team consisting of Karan Varindani, Wenyang Zhang, and Sibo Zhu for our Machine Learning final project (Boston University CS 542, Summer 2017).

The challenge

Our project is based on research started by HAL on learning a convolutional neural network for non-uniform motion blur removal.

Our approach

Our project approach is to estimate motion blur from a single, blurry image. We propose a deep learning approach to predict the probabilistic distribution of motion blur at the patch level using a Convolutional Neural Network (CNN).

Credits

• Core algorithm: Sibo Zhu
• Data creation: Wenyang Zhang
• Project management: Karan Varindani

---

You can find our Code here, and our report here.