COVID-19 Research @ CSE IIT Bombay
-
When will the Covid-19 Pandemic End?
Levitt’s Metric on Indian Data
Bhaskaran Raman, Dept of CSE, IIT Bombay
Link: https://bit.ly/levitt-india
-
Tapestry
Compressed sensing is a paradigm for fast image acquisition, coupled with clever decoding techniques. The application of this for pooled testing is somewhat surprising. The broad idea is that you consider a vector 'x' of n elements, each containing the viral concentration in the naso/oro-pharyngeal swab of a particular patient. A typical vector 'x' in many useful scenarios is essentially sparse since most people are not infected. A technician pools (i.e. mixes)together small but equal portions of the samples of small subsets of these n people. Let us say you create m << n pools. Then you test only these m poolsin the RTPCR machine instead of testing all n people. Let 'y' be a vector of the viral concentrations in the m pools. Then you can write y = Ax + noise, where A denotes the 'mixing/pooling matrix' telling you which samples went into which pool. Given y and A, you have to infer 'x'. There are interesting techniques in compressed sensing theory which tell you how to create a "good" matrix 'A', i.e. which allows for accurate reconstruction of sparse vectors 'x'. Of course, there are some constraints unique to this problem, such as the requirement that A be binary as well as sparse. In our design, a given sample contributes to no more than 3 pools. Our work attempts to estimate the viral concentrations apart from telling you which patients were infected.
Drafts of the project have beein submitted to MEDARXIV and ARXIV
Contributors :-
Faculty- Prof. Ajit Rajwade
PHD student- Sabyasachi Ghosh
Undergrad students- Rishi Agarwal, Mohammad Ali Rehan, Shreya Pathak, Pratyush Agarwal, Yash Gupta, Sarthak Consul, Nimay Gupta, Ritika Goyal
-
Contact-free exam management
SAFE for online exams (contact-free exam) created by Prof. Bhaskaran Raman and Prof. Kameswari Chebrolu
-
E-Token systems to avoid waiting/queues/crowding (Link)
In Partnership with ISRC (Indian Scientists’ Response to CoViD-19)
Faculty in-charge:- Prof. Bhaskaran Raman
-
Non-Covid Hospitals
Technical partner is Project Mumbai's initiative for Non-Covid Hospitals and In Partnership with ISRC (Indian Scientists’ Response to CoViD-19)
Faculty in-charge:- Prof. Bhaskaran Raman
-
APRICOT
A PRIvacy preserving COntact Tracing system
Released under open source (Link)
In Partnership with ISRC (Indian Scientists’ Response to CoViD-19)
Faculty in-charge:- Prof. Bhaskaran Raman
-
SAFE for Checking Quarantine Adherence (Link)
Faculty in-charge:- Prof. Bhaskaran Raman and Prof. Kameswari Chebrolu
-
WWH Project
The Project is being used by KEM hospital and IITB hospital for tele-medicine purposes.
Faculty in-charge:- Prof. Kameswari Chebrolu
-
BodhiTree (Link)
A learning management system for teachers to go online. This platform is used to conduct a massive workshop on online teaching which saw 10,000 registrations. Many teachers (700+) have expressed interest on using the platform for their teaching going forward. (The workshop details are already on CSE website)
Faculty in-charge:- Prof. Kameswari Chebrolu
-
Kaawakate platform(Link)
Helps journalists fact check misinformation. Currently in talks with fact checkers (like viacom, factly) to adopt the platform. Click here for more details.
Faculty in-charge:- Prof. Kameswari Chebrolu
-
CORONTINE: Tracking and tracing of asymptotic carriers during pandemic(Link)
A evolved common Logic for Red Zone Violation Detection and Possible Flowchart based on VLR/CDR within TSP is developed. Click here for more details.
We built robust models for anomaly detection at the TSP and shared code with the individual TSPs. Click here for more details.
Corintine has been integrated with IoT. Check this video.
Faculty in-charge:- Prof. Ganesh Ramakrishnan
-
Contactless (video) surveillance at COVID-19 quarantine facilities
Background knowledge about this initiative can be read in this Insight article.
After the initial successful use of the contactless survelliance system at the Vanvihar guest house,
The system was used for:-- Crowd counting to ensure that there is no congregation of more than 2-3 people in the campus(IITB) at select locations
- Face recognition to ensure that close to 150 people identified by the institute who are not supposed to go outside the institute indeed remain within the institute.
- ANPR (automatic number plate recognition) enabled on specific cameras to ensure that only permitted vehicles enter and exit the campus(IITB).
Faculty in-charge:- Prof. Ganesh Ramakrishnan
-
Lokacart
Background knowledge about this initiative can be read in this Insight article.
With the help of our industry partners, this is being actively taken forward with Kirana stores, more farmer groups, organizations such as BAIF (Bharat Agro Industrial Foundation that is working with 4 lakh farmers). With a help of a team that is much well organized now, we are also carefully working out plans for more systematic and professional adoption. Lots of field level activity and collaborative efforts are ongoing.
Faculty in-charge:- Prof. Ganesh Ramakrishnan
-
Lokavidya: Knowledge sharing platform for educational institutions
Background knowledge about this initiative can be read in this Insight article
With the help of our industry partners, this is being actively evolved/developed for Schools and skill development organizations as well as farmer organizations such as BAIF (Bharat Agro Industrial Foundation) that is working with 4 lakh farmers). With some of the entities, we are working on need assessment and further refining of our software. We are evolving proposals for joint collaboration.
Faculty in-charge:- Prof. Ganesh Ramakrishnan