Real Time Engagement Analysis
USC, LA
Project: https://github.com/anchalkpr/Audience_Engagement_Analysis

The use of computer vision to analyze both environmental contexts and affect indicators is an emergent field to which all of us were immediately attracted. Our interests evolved from an exploration of context and posture capture to a more speci fic and functional emphasis on monitoring and analyzing the perceived levels of engagement in a lecture audience, for the bene fit of the orator. As engagement itself is immeasurable, our approach necessarily centers on the behavioral indicators of this activation modality, consequently presuming that the external states of audience members faithfully correlate with their internal cognitive activation.

Automatic Text Summarization of Hindi Articles
USC, Los Angeles
Project: https://github.com/USC-NLP-17/NBNewSummarizer

Information on the web in the form of text and documents is enormous and is growing at an exponential pace. By providing relevant information in the form of summaries or abstracts using most informative sentences in a coherent way, automatic summarization becomes an important task in order to deal with swiftly increasing digital data. Though there is abundant information existing in Hindi, not much work has been done on Hindi text summarization. In this project we focus on unsupervised learning techniques to generate extractive summaries from Hindi texts.

Live Traffic Interactive Map
USC, LA
Project: http://amit-jha.com/livebus/

A live portal dedicated to track real-time traffic movement in the bay area with an interactive map, developed using PostGIS, AngularJS, D3JS and GCP. The portal provides features to filter and track subsets of routes. This can be extended to other cities.