search

kob51 / image_deblurring

Implementation of "Image Deblurring Using Inertial Measurement Sensors" for the Computational Photography final project (Fall 2020)
8 stars 4 forks source link

Could you provide some data #1

Open zlthaha opened 1 year ago

zlthaha commented 1 year ago

I wonder if you could provide some data,I'd like to test it

kob51 commented 1 year ago

I used data from the publicly-available ETH3D datasets, specifically "plant_5" and "camera_shake_1"

https://www.eth3d.net/slam_datasets

zlthaha commented 1 year ago

Thank you for providing the dataset URL!

I have a few questions about the code and the paper:

  1. You have selected a blur core of size 35. Does it mean that you use the same blur core for the whole world? However, it is mentioned in the paper that this method uses spatially varying fuzzy kernel.

  2. Did you perform well in the test? I seem to have gotten a bad result.

Thanks again!

kob51 commented 1 year ago

I've attached my report with further implementation details and results Image_Deblurring_Using_Inertial_Measurement_Sensors.pdf

  1. The size of the kernel is constant but as you will see from the black/white figures in the report, a different kernel is used in different parts of the image
  2. It wasn't perfect, but on the plant datasets there were definitely some good results. The camera shake dataset had extreme motion which this algorithm performed poorly on, however.