Me talking

Captioning images using Tensorflow

30 November 2019

This talk was based mostly on my past exploration and testing the limits of Tensorflow.js . Prior work was based on training small models in the browser or porting Tensorflow models to working with TF.js in the browser. This time I tried tweaking the larger Tensorflow Image Caption models to be optimized and ported for browser. I couldn’t get it work properly. So I presented my learning about how TF.js is useful and what real world restrictions (available data-plan, device capabilities) come into play when doing large tasks like Image Captioning in browser.

The audience was very interactive and participated with curious questions. They also had some questions about how to become a GDE :-) I loved the interaction <3