**This page is heavily outdated. Last updated on March 2013**

IVR Junction [@ Microsoft Research]
Project Website: http://www.ivrjunction.org
Advisor: Dr. Bill Thies
Publication: Aditya Vashistha and William Thies, "IVR Junction: Building Scalable and Distributed Voice Applications in the Developing World", in 6th USENIX/ACM Workshop on Networked Systems for Developing Regions (NSDR 2012), June 2012, Boston, US.
Interactive Voice Response (IVR) systems play an important role in collecting and disseminating information in developing regions. Recently, researchers have used IVR technology to build voice forums, in which callers leave messages that can be heard over the Internet and over the phone. However, despite their appeal, voice forums remain difficult to set up, and difficult to scale due to the overhead of moderating content and the cost of phone calls.
We present IVR Junction: an IVR platform that leverages existing free services and commercial tools to simplify the process of creating a voice forum. IVR Junction utilizes familiar cloud-based services to provide free content hosting and moderation, as well as a novel mechanism for automatically synchronizing content across geographically-dispersed offices, thereby enabling local access points with decreased calling costs.
Project Website: http://www.ivrjunction.org
Advisor: Dr. Bill Thies
Publication: Aditya Vashistha and William Thies, "IVR Junction: Building Scalable and Distributed Voice Applications in the Developing World", in 6th USENIX/ACM Workshop on Networked Systems for Developing Regions (NSDR 2012), June 2012, Boston, US.
Interactive Voice Response (IVR) systems play an important role in collecting and disseminating information in developing regions. Recently, researchers have used IVR technology to build voice forums, in which callers leave messages that can be heard over the Internet and over the phone. However, despite their appeal, voice forums remain difficult to set up, and difficult to scale due to the overhead of moderating content and the cost of phone calls.
We present IVR Junction: an IVR platform that leverages existing free services and commercial tools to simplify the process of creating a voice forum. IVR Junction utilizes familiar cloud-based services to provide free content hosting and moderation, as well as a novel mechanism for automatically synchronizing content across geographically-dispersed offices, thereby enabling local access points with decreased calling costs.

Incentivizing Viral Spread of Micro-Tasks
Advisor: Dr. Bill Thies
The project investigates how different models of financial incentives make a micro-task go viral.
Advisor: Dr. Bill Thies
The project investigates how different models of financial incentives make a micro-task go viral.

Mobilizing Crowd in India [@ Microsoft Research]
Project Website: http://whodunitchallenge.com
Advisor: Dr. Bill Thies
The project investigates how people in the developing world harness existing ICTs and communication resources for real time collaboration, to solve time critical problems, for urgent mobilization and for tasks that require wide-area team building. The project is analogous to DARPA Network Challenge.
Project Website: http://whodunitchallenge.com
Advisor: Dr. Bill Thies
The project investigates how people in the developing world harness existing ICTs and communication resources for real time collaboration, to solve time critical problems, for urgent mobilization and for tasks that require wide-area team building. The project is analogous to DARPA Network Challenge.

Usable IVR Systems [@ Microsoft Research]
Advisor: Dr. Bill Thies
We are adding two visual components in IVR systems to make them more usable. The first component enables IVR service providers to visualize call data on a mapping platform like Ushahidi, Google Maps etc. by using an open source API which can be interfaced with any IVR platform and any mapping platform. This helps them to have birds-eye-view of how people are interacting with their IVR application for disaster relief efforts, grievance redressal etc. Location data also enables IVR systems to provide location based information to IVR callers.
The second component aims to improve urban users’ IVR experience by visualizing IVR call tree on a smart phone as soon as they call an IVR application from a nearby second phone and put the phone on a speaker mode. We are embedding call tree data in the audio stream, which is then decoded and visualized by the smart phone application. The visualized call tree results in easy navigation of IVR applications’ tedious and arcane call trees.
Advisor: Dr. Bill Thies
We are adding two visual components in IVR systems to make them more usable. The first component enables IVR service providers to visualize call data on a mapping platform like Ushahidi, Google Maps etc. by using an open source API which can be interfaced with any IVR platform and any mapping platform. This helps them to have birds-eye-view of how people are interacting with their IVR application for disaster relief efforts, grievance redressal etc. Location data also enables IVR systems to provide location based information to IVR callers.
The second component aims to improve urban users’ IVR experience by visualizing IVR call tree on a smart phone as soon as they call an IVR application from a nearby second phone and put the phone on a speaker mode. We are embedding call tree data in the audio stream, which is then decoded and visualized by the smart phone application. The visualized call tree results in easy navigation of IVR applications’ tedious and arcane call trees.

Talent Hunt: Community Moderated IVR Systems [@ Microsoft Research]
Project Website: http://research.microsoft.com/talenthunt
Advisor: Dr. Bill Thies
In recent years, IVR systems have spanned diverse domains, including citizennews journalism,agriculturaldiscussion forums, communitydialogue, user-generatedmaps, access tohealth information, outreachto sex workers, groupmessaging, feedback on school meals, supportfor community radio stations, ruralemployment exchange, and viral entertainment platform. Despite the enthusiasm around them and evidence that proves that IVR applications are having the envisioned impact, the biggest challenge in scaling Interactive Voice Response (IVR) systems is to figure out how to scale moderation of content.
Community moderation has played pivotal role in success of web 2.0 sites like stackoverflow.com, quora.com, Yahoo! Answers etc. The research goal of Talent Hunt project is to design community moderation algorithms for IVR systems, evaluate usability of community moderation voice user interface for several communities in the developing world, and investigate efficacy of community moderation to scale IVR systems. We are also investigating that if the content is appealing, would people be willing to pay to use the system.
In order to that, we have designed an IVR application in entertainment domain,Talent Hunt, where callers can record a voice talent (song, poem, joke, one-liner etc.), listen and rate talents recorded by others, and share talents with their friends. So far our application is called 4900 times and the total time spent by callers is more than 12860 minutes.
Project Website: http://research.microsoft.com/talenthunt
Advisor: Dr. Bill Thies
In recent years, IVR systems have spanned diverse domains, including citizennews journalism,agriculturaldiscussion forums, communitydialogue, user-generatedmaps, access tohealth information, outreachto sex workers, groupmessaging, feedback on school meals, supportfor community radio stations, ruralemployment exchange, and viral entertainment platform. Despite the enthusiasm around them and evidence that proves that IVR applications are having the envisioned impact, the biggest challenge in scaling Interactive Voice Response (IVR) systems is to figure out how to scale moderation of content.
Community moderation has played pivotal role in success of web 2.0 sites like stackoverflow.com, quora.com, Yahoo! Answers etc. The research goal of Talent Hunt project is to design community moderation algorithms for IVR systems, evaluate usability of community moderation voice user interface for several communities in the developing world, and investigate efficacy of community moderation to scale IVR systems. We are also investigating that if the content is appealing, would people be willing to pay to use the system.
In order to that, we have designed an IVR application in entertainment domain,Talent Hunt, where callers can record a voice talent (song, poem, joke, one-liner etc.), listen and rate talents recorded by others, and share talents with their friends. So far our application is called 4900 times and the total time spent by callers is more than 12860 minutes.

Voice Based Social Media and Information Delivery System for Farmers [as a Fellow of NIXI]
Pilot Deployment: Pilot deployment and field study is facilitated by Enzen in Devanahalli, Bangalore
Advisor: Dr. Rajarathnam Nallusamy
Publication: Aditya Vashistha and Rajarathnam Nallusamy, "Voice Based Social Media and Information Delivery System for Farmers" in 2012 Workshop on Mobile and Web Technologies in Social and Economic Development.
Farmers face various problems which results in low productivity, lower profits and higher indebtedness which eventually lead some of them to commit suicide. There is a need to find solutions to eliminate or minimize interference of middlemen, to effectively disseminate knowledge among farmers, to bridge the gap between farmer, peer farmers, agriculture extension workers and experts, and to enable farmers financially. However, there are various challenges and constraints which an ICT system needs to address. Many farmers have low or no literacy level. They have limited financial resources and limited availability of electricity. Only device which most farmers use is a low cost 2G mobile phone but most of them do not compose or read text message. As voice is a natural and accessible medium for farmers having limited literacy level, we use a voice based social network and information delivery system which:
Pilot Deployment: Pilot deployment and field study is facilitated by Enzen in Devanahalli, Bangalore
Advisor: Dr. Rajarathnam Nallusamy
Publication: Aditya Vashistha and Rajarathnam Nallusamy, "Voice Based Social Media and Information Delivery System for Farmers" in 2012 Workshop on Mobile and Web Technologies in Social and Economic Development.
Farmers face various problems which results in low productivity, lower profits and higher indebtedness which eventually lead some of them to commit suicide. There is a need to find solutions to eliminate or minimize interference of middlemen, to effectively disseminate knowledge among farmers, to bridge the gap between farmer, peer farmers, agriculture extension workers and experts, and to enable farmers financially. However, there are various challenges and constraints which an ICT system needs to address. Many farmers have low or no literacy level. They have limited financial resources and limited availability of electricity. Only device which most farmers use is a low cost 2G mobile phone but most of them do not compose or read text message. As voice is a natural and accessible medium for farmers having limited literacy level, we use a voice based social network and information delivery system which:
- Minimize dependence of farmer on middlemen by informing farmers the current market price of their produce.
- Eliminate middlemen by providing a web based platform, using the existing infrastructure and low end devices, to connect technically impaired farmers to technical savvy major food service providers or bulk buyers.
- Disseminate knowledge about modern agricultural practices and other technical innovations in spite of illiteracy barrier.
- Provide a voice based social network for farmers to interact with peer farmers, agriculture extension workers and agriculture experts.
- Provide a platform for Micro-finance institutions to connect to farmers.

NoMark: Digital Video Copyright Protection System [@ Infosys Labs]
Advisor: Dr. Rajarathnam Nallusamy and Dr. Sanjoy Paul
Publications:
Patent: Aditya Vashistha, Rajarathnam Nallusamy and Sanjoy Paul, "Method and System for Efficient Watermarking of Digital Video” filed for Indian Patent (Application No: 3779/CHE/2010) and US patent (Application No: 13/027,675).
NoMark project is a POC system for a filed patent application which protects the digital video content against copyright infringement, piracy and illegal distribution without embedding any watermark information. In addition, copyright protection is extended on each scene of the NoMark protected video.
Usually, watermarking algorithms embed watermark data into an image or a video in an invisible or visible manner. These algorithms are prone to many known watermark attacks aimed at detecting and distorting the embedded data. As no data is embedded into the NoMark protected video, the video is not susceptible to many traditional watermarking attacks. NoMark method is robust, novel and imperceptible. The method is robust against a wide variety of image and video watermarking attacks.
Advisor: Dr. Rajarathnam Nallusamy and Dr. Sanjoy Paul
Publications:
- Aditya Vashistha, Rajarathnam Nallusamy, Amitabha Das and Sanjoy Paul, “Watermarking Video Content Using Visual Cryptography and Scene Averaged Image”, in International Workshop on Content Protection and Forensics (CPAF 2010), in proceedings of IEEE International Conference on Multimedia and Expo (ICME 2010), July 19 - 23, 2010, Singapore.
- Aditya Vashistha, Rajarathnam Nallusamy and Sanjoy Paul, “NoMark: A Novel Method for Copyright Protection of Digital Videos Without Embedding Data”, in The IEEE International Symposium on Multimedia (ISM 2010), December 13-15, 2010, Taichung, Taiwan.
Patent: Aditya Vashistha, Rajarathnam Nallusamy and Sanjoy Paul, "Method and System for Efficient Watermarking of Digital Video” filed for Indian Patent (Application No: 3779/CHE/2010) and US patent (Application No: 13/027,675).
NoMark project is a POC system for a filed patent application which protects the digital video content against copyright infringement, piracy and illegal distribution without embedding any watermark information. In addition, copyright protection is extended on each scene of the NoMark protected video.
Usually, watermarking algorithms embed watermark data into an image or a video in an invisible or visible manner. These algorithms are prone to many known watermark attacks aimed at detecting and distorting the embedded data. As no data is embedded into the NoMark protected video, the video is not susceptible to many traditional watermarking attacks. NoMark method is robust, novel and imperceptible. The method is robust against a wide variety of image and video watermarking attacks.

2PASCD: 2 Pass Abrupt Scene Change Detection Algorithm [@ Infosys Labs]
Advisor: Dr. Rajarathnam Nallusamy and Dr. Sanjoy Paul
Publication: Aditya Vashistha, Rajarathnam Nallusamy and Sanjoy Paul, “2PASCD: An efficient 2-Pass Abrupt Scene Change Detection Algorithm”, in 2010 International Conference on Multimedia Information Networking and Security (MINES 2010), IEEE Press, November 4-6, 2010, Nanjing, China.
Patent: Aditya Vashistha, Rajarathnam Nallusamy and Sanjoy Paul, "Methods, Systems and Computer-Readable Media for Detecting Scene Changes in a Video”, filed for US Patent (Application No: 11/328,308).
Scene change detection is instantly performed by human but vast computational resources and complex algorithms are required to automate the process. An efficient 2 pass abrupt scene change detection algorithm is incubated that results in high precision and recall in optimal time. The efficiency of scene change detection is improved by a second level inspection of scene changes detected after first pass. Second level inspection is done using a fixed average and a varying average to ascertain whether the scene boundary detected is correct or the scenes need to be combined. False positives are removed to improve the precision values. The algorithm is fast as it is based on color histograms and results in average precision and recall values higher than 95%.
Advisor: Dr. Rajarathnam Nallusamy and Dr. Sanjoy Paul
Publication: Aditya Vashistha, Rajarathnam Nallusamy and Sanjoy Paul, “2PASCD: An efficient 2-Pass Abrupt Scene Change Detection Algorithm”, in 2010 International Conference on Multimedia Information Networking and Security (MINES 2010), IEEE Press, November 4-6, 2010, Nanjing, China.
Patent: Aditya Vashistha, Rajarathnam Nallusamy and Sanjoy Paul, "Methods, Systems and Computer-Readable Media for Detecting Scene Changes in a Video”, filed for US Patent (Application No: 11/328,308).
Scene change detection is instantly performed by human but vast computational resources and complex algorithms are required to automate the process. An efficient 2 pass abrupt scene change detection algorithm is incubated that results in high precision and recall in optimal time. The efficiency of scene change detection is improved by a second level inspection of scene changes detected after first pass. Second level inspection is done using a fixed average and a varying average to ascertain whether the scene boundary detected is correct or the scenes need to be combined. False positives are removed to improve the precision values. The algorithm is fast as it is based on color histograms and results in average precision and recall values higher than 95%.

LongArm: Remote Mobile Phone Application System [@ Infosys Labs]
Advisor: Dr. Arun Agrahara Somasundra
Publication: Karthikeyan Balaji Dhanapal, K Sai Deepak, Saurabh Sharma,Sagar Prakash Joglekar, Aditya Narang, Aditya Vashistha, Paras Salunkhe, Harikrishna G. N. Rai, Arun Agrahara Somasundara and Sanjoy Paul, “An Innovative System for Remote and Automated Testing of Mobile Phone Applications”, in Annual SRII Global Conference 2012, July 2012, San Jose, US.
LongArm is a robotic system for remote testing of mobile and IPTV applications. In this system, the handset is in the cellular network service area but the tester is in a remote location.
The tester controls the handset over the Internet using LongArm. The system does not require any changes to be done to the mobile phone itself. In LongArm the input given by the human tester is replaced by a 3-axis robot which does the pressing of the keys. The output is captured by a camera. This robot-camera system is placed in the location where the phone network is available (say, US), and controlled from a remote location (say, India). This possibility gives opportunities to leverage the potential of the global outsourcing business model. In addition, the system is agnostic to the Operating System & application running on the mobile phone, and is also non-intrusive.
Advisor: Dr. Arun Agrahara Somasundra
Publication: Karthikeyan Balaji Dhanapal, K Sai Deepak, Saurabh Sharma,Sagar Prakash Joglekar, Aditya Narang, Aditya Vashistha, Paras Salunkhe, Harikrishna G. N. Rai, Arun Agrahara Somasundara and Sanjoy Paul, “An Innovative System for Remote and Automated Testing of Mobile Phone Applications”, in Annual SRII Global Conference 2012, July 2012, San Jose, US.
LongArm is a robotic system for remote testing of mobile and IPTV applications. In this system, the handset is in the cellular network service area but the tester is in a remote location.
The tester controls the handset over the Internet using LongArm. The system does not require any changes to be done to the mobile phone itself. In LongArm the input given by the human tester is replaced by a 3-axis robot which does the pressing of the keys. The output is captured by a camera. This robot-camera system is placed in the location where the phone network is available (say, US), and controlled from a remote location (say, India). This possibility gives opportunities to leverage the potential of the global outsourcing business model. In addition, the system is agnostic to the Operating System & application running on the mobile phone, and is also non-intrusive.

PalmPal: Enabling Software Engineering in Mobile Devices [as Intern @ Infosys]
Language: J2ME Tool: NetBeans IDE
This project is a POC system for a filed patent application which enables the users to access the Software engineering tools like use case diagram, deployment diagram, activity diagram, class diagram, ER diagram on mobile. User can generate the equivalent java code from class diagram and can generate SQL queries for oracle database from ER diagram using forward engineering approach. User can also create the class diagram from a java code snippet using backward engineering. Also the diagrams can be converted into svg files for desktop supported format.
Language: J2ME Tool: NetBeans IDE
This project is a POC system for a filed patent application which enables the users to access the Software engineering tools like use case diagram, deployment diagram, activity diagram, class diagram, ER diagram on mobile. User can generate the equivalent java code from class diagram and can generate SQL queries for oracle database from ER diagram using forward engineering approach. User can also create the class diagram from a java code snippet using backward engineering. Also the diagrams can be converted into svg files for desktop supported format.