RNNs in Computer Vision

Agenda:

• RNNs in Computer Vision by Mr. Ravi Kiran Sarvadevabhatla (2 hrs)

• Open discussion (0.5 hr)

About the talk

The talk will cover the below topics

1. Motivation for Recurrent Neural Networks (RNNs)

2. RNN - usage paradigms, formulation, learning

3. Drawbacks of RNN, motivation and design evolution for LSTM

4. RNN/LSTM-based architectures in computer vision

5. RNN vs CNN revisited, RNNs v/s HMM

6. A case study: RNNs for on-line recognition of hand-drawn sketches

About the speaker

Ravi is a PhD student at IISc (Video Analytics Lab). Before joining IISc, he worked as a Research Engineer at Duos Technologies, USA (http://www.duostechnologies.com/) from May 2012 to December 2013. Prior to that, he was a Research Engineer at Honda Research Institute USA (http://usa.honda-ri.com/) from January 2008 to April 2012. He completed his masters in Computer Science (http://www.cs.washington.edu) from the University of Washington (http://cds.iisc.ac.in/student/ravikiran/www.washington.edu) in 2008. He got an integrated MS/B.Tech degree (Honors) in Computer Science from IIIT-H (https://www.iiit.ac.in/) in 2004.

*No Attending charges.

Participants are strongly encouraged to bring their laptops. While we do not do any programming during the meetup, we have found it to be quite handy to go through portions of the week's topic and to make notes and discuss.

IMPORTANT NOTE: Please sign up only if you can attend, since we have space restrictions. Starting Jan 2017, members who sign-up but do not turn up for 3 meetups will be blocked. So kindly check your schedule before signing up.

Please find the meetup location by following this link (https://www.google.co.in/maps/place/Soliton+Technologies+Pvt+Ltd/@12.906443,77.591944,17z/data=%213m1%214b1%214m5%213m4%211s0x3bae1510e837c5f5:0x8bfee78fb30fc8c9%218m2%213d12.906443%214d77.594138).