Joint SIG Mtg: Rock Physics and Data Science and Machine Learning - Apr 5th

Complete Title: Unsupervised Segmentation of CT Images Using Deep Learning


This month's meeting will be a joint meeting of the Rock Physics SIG and the Data Science and Machine Learning SIG

This is a Hybrid Event, you will still need to pre-register for either in person or online.

NOTE: You Must Be Logged In to Register.

Meeting Location:
GSH Geoscience Center & Museum 
1790 W. Sam Houston Parkway N
Houston TX 77043

Speaker: Jyoti Behura, Founder, Seismic Science LLC

The most common method for segmenting CT images in digital rock physics relies on supervised learning, where the user employs labeled images. Each pixel of a CT image is classified the user as a grain (or type of grain) or a pore. Although supervised segmentation performs satisfactorily, there are multiple drawbacks to this approach that include need for user input, introduction of user bias, time consuming process, need for independent ground truth, and poor generalization of the segmentation models. Here, I introduce a methodology to overcome the above challenges by developing an unsupervised segmentation technique that does not rely on user input.

Speaker Biography: Jyoti Behura, Founder Seismic Science LLC
Dr. Jyoti Behura is a research assistant professor in the Geophysics Department. He is also the founder of Seismic Science LLC, a technology start-up developing seismic imaging technologies for hydraulic fracturing, CCS, and geothermal applications. Prior to that he was part of the imaging team at BP Americas Inc. His primary research interests revolve around imaging and full-waveform inversion. He, along with his PhD advisor, Ilya Tsvankin, received the 2009 best paper in Geophysics Award. He is also the recipient of the J. Clarence Karcher Award in 2012.

 **Access information will be sent to all registrants after registration closes.

Price List:

Pre-Registered Late/Walk-Up

In Person or Virtual

Free Free
In Person or Virtual
$10 $10

** Please be aware that ALL registrations must go through the 'Checkout', even if your total is zero. Registration is not confirmed until you complete the checkout process.



4/5/2023 5:15 PM - 6:30 PM
Central Daylight Time

Sign In