Name: TriPython September 2019 Meeting: Deep Processing of Ridiculously Large Images
Start: 2019-09-26T19:00:00-04:00
End: 2019-09-26T21:00:00-04:00
Location: Caktus Group

To get full and timely notifications of all TriPython events and discussions, subscribe to the TriPython email list at

http://mail.python.org/mailman/listinfo/trizpug/

Full information about TriPython is available on the website at

http://tripython.org

Object detection algorithms with convolutional neural networks (CNNs) are typically applied to small images, usually smaller than 1000 x 1000 pixels. If an image is larger than this, often it will be scaled down to the "right" size, but what if scaling your images down would erase the most crucial details? Because of these concerns, our speaker recently has been adapting CNN-based object detection algorithms to data acquired from images that are on the order of 10k x 10k pixels (100 megapixels).

In this talk, Dr. Eric Thomson will give an introduction to new imaging technology he has been working on to acquire this data (basically it is a compound eye like you find in the insect world, but made out of cameras). He will then give a quick beginner-friendly overview of neural networks, deep learning, and CNNs. Then, he will describe the tensorflow-based pipeline he has developed to automatically annotate the ridiculously large images that we acquire using the new tech.

Dr. Thomson is a Research Scientist at Duke Neurobiology. Dr. Thomson currently works on a large-scale imaging system that lets him collect data from lots of cameras, stitch them together, and then analyze the data. He also hosts TriPython's monthly Durham Project Night at Caktus Group.

Park in the municipal deck on the other side of the Arts Council across W. Morgan St. The after-meeting will be around the corner at Bull McCabe's Irish Pub. Come join us for a fun and informative evening.

Chris Calloway

Triangle Python Users Group

Technology

Python

Triangle Python Users Group

TriPython September 2019 Meeting: Deep Processing of Ridiculously Large Images

Triangle Python Users Group

Details

Members are also interested in