• Hello Visitor!

    I am a masters student in Informatics with a focus on Deep Learning applications in Computer Vision and Biomedical Imaging.

    Research Interests : Transfer Learning, Domain Adaptation, Few-Shot Learning, Computer Vision, Image/Video Analysis

    View CV GitHub LinkedIN Twitter

About Me

Hi I'm Shivangi. Currently, I am pursuing Masters In Informatics (Computer Science) at TUM, specializing in Computer Graphics and Vision Chair at the university. I hold a Bachelor's Degree In Computer Science from NIT Hamirpur, India. In the past semesters, I have done a couple of deep learning projects related to vision and biomedicine (check out my CV/GitHub for details). This semester I am working towards my Master Thesis under the guidance of Prof. Matthias from the Visual Computing Lab at the university. My work mainly focuses on transfer learning for forgery detection in videos.

Invited Events

I was invited to the following events

Eastern European Machine Learning Summer School (EEML), Bucharest, Romaina

Selected as one of the 150 participants (out of 700+ applicants) to attend the EEML Summer School on Deep Learning and Reinforcement Learning. This was held in Bucharest, Romania from 1-6 July 2019. Lectures and hands-on practical sessions were given by one of the best researchers in the domain. I was also awarded full scholarship to attend the school.

Advanced Technology Higher Education Network(ATHENS) Programme at TU Delft, Netherlands

Selected as one of the students to attend the ATHENS exchange programme at one of the parter universities. It is a 7-day exchange programme to facilitate exchanges between students of the major European technological institutions. I was selected for the course Finite Element Algorithms at TU Delft. It was held from 16-23 March 2019. I was awarded full travel grant for the programme.

Research Projects

Multi Modal Brain Structure Segmentation With Adversarial Learning

A Guided Research where the 3D Multi-modal (FLAIR and T1) brain MRI scans were used in a semi-supervised setting to perform semantic segmentation. The results were slightly improved by leveraging the unannotated scans. GANs were used for this task.

Polyp Localization in Colonoscopy Videos

A Machine Learning in Medical Imaging (MLMI) practical course where the aim was to detect polyps in the image. First, the results were evaluated on standard benchmark CVC-Polyp dataset (which consits of colonoscopy videos from different studies), and then on the dataset provided by the hospital. For this project, SSD (Single Shot Multibox Detector) was used. Presented this work during poster session at EEML Summer School.

Deep Image Composting

A (team)project as part of my Advanced Deep Learning for Computer Vision course where we could build a model that creates realistic composite images. Conditional GANs along with specialized loss for the generator were used to achieve this task.

Deep Clustering

A practical course by Computer Vision chair wherein my project was to improve clustering accuracy on various toy datasets like MNIST, CIFAR-10, and STL. I experimented with autoencoders for this.

Cooperative Spectrum Sensing in Cognitive Radio

Secondary users could use spectral band when not used by primary users. Local and cooperative sensing was used for energy detection.



Pursuing MS in Informatics (Specializating in Computer Vision and Deep Learning)

Courses Undertaken: Introduction To Deep Learning, Advanced Deep Learning for Computer Vision, Machine Learning, Machine Learning for Computer Vision, Machine Learning In Medical Imaging Practical Course, Natural Language Processing Seminar, Deep Learning for Computer Vision and Biomedicine Practical Course, Protein Prediction For Computer Scientists

Awarded Gold Medal for graduating top of Computer Science Batch (2011-2015)

Courses Undertaken: Analysis & Design of Algorithms, Probability and Queuing Models, Data Structures, Discrete Structure, Differential and Integral Calculus, Operating System, Theory Of Computation, Computer Networks, Data Mining and Knowledge Discovery

Worked as an active member of college society GLUG (GNU Linux Users Group), C-SEC (College Departmental Society) throughout all semesters. Literacy Mission: Taught the underprivileged students for two semesters. Team .exe: Worked as Coordinator of the team during Nimbus 2014, the Annual Tech Fest of NIT Hamirpur. Fine Arts: Worked as Co-Convener during Hill'ffair 2013 , the Annual Cultural Fest of NIT Hamirpur. Public Relation Club: Worked as member of team during Hill'ffair 2011.

Studied Mathematics, Chemistry and Physics.

  • Graduated with 94% marks.
  • Awarded Roll Of Honour for graduating top of the High School (Science and Mathematics).

Work Experience

Machine Learning WerkStudent at Siemens Feb 2019 - Present

  • Designed a visualizer for analyzing the error ( difference between Ground Truth and Predicted) GPS coordinates.
  • Used of Kalman Filters for the purpose of noise and jitter removal from the sensor data.
  • Building models for activity prediction (Walk, Run, Car, Bus, Tram, Train etc) of the users.

Software Engineer at Nucleus Softwares June 2015 – Aug 2017

  • Algorithms Implemented: K-Means Clustering , Association Rule Mining, Logistic Regression, Multiple Linear Regression, Sequence Mining in Java 8 for Lending Analytics Product
  • Technical Upgradation: From Struts 1.x and Java 6 to Spring Framework and Java 8
  • Performance Upgrade: Improved Model Building Time of Decision Tree algorithm
  • Webservices: Developed using REST API
  • Module Implemented:Data Preprocessing
  • Three-month rigorous training for on Banking/Lending fundamentals
  • Knowledge Transfer Sessions for the understanding of Lending Analytics Product
  • Bug Fixing (Frontend/Backend Related) for three months
  • Database Script: (Creation, Rollback and Cleanup) for the entire application

Summer Intern at IIT Bombay May 2014 - Jun 2014

  • AAQ (Ask A Question) Forum : Developed a forum for interaction between IIT professors and students
  • Also developed an android application for same forum to provide inter-connectivity with hand-held devices