Sonal Bihani

Sonal Bihani

Waterloo, Ontario, Canada
2K followers 500+ connections

Join to view profile

About

I am a Data Science Intern at Altair, a global leader in data intelligence solutions…

Activity

Join now to see all activity

Experience

  • Clio - Cloud-Based Legal Technology Graphic

    Clio - Cloud-Based Legal Technology

    Toronto, Ontario, Canada

  • -

  • -

    Toronto, Ontario, Canada

  • -

    Hyderabad, Telangana, India

  • -

    Hyderabad, Telangana, India

  • -

    Hyderabad, Telangana, India

  • -

    Hyderabad Area, India

  • -

    Vellore, Tamil Nadu, India

  • -

    Kolkata, West Bengal, India

  • -

  • -

    Kolkata, West Bengal, India

  • -

Education

Licenses & Certifications

Join now to see all certifications

Courses

  • Computer Vision

    CS 684

  • Data Intensive Distributed Computing

    CS 651

  • Data Visualization

    STAT 842

  • Deep Learning

    STAT 940

  • Exploratory Data Analysis

    STAT 847

  • Machine Learning

    CS 680

  • Neural Networks

    CS 679

  • Optimization for Data Science

    CO 673

  • Statistics for Data Science

    STAT 847

Projects

  • Semantic Segmentation using Deep Learning for Medical Image Analysis

    -

    The objective of this project was to implement and explore various components of a deep learning pipeline for semantic segmentation. Semantic segmentation involves classifying each pixel in an image into specific categories, enabling precise identification and localization of structures or abnormalities. The project addresses a critical aspect of medical image analysis, where precise segmentation is pivotal for accurate diagnosis and treatment planning. The implemented techniques contribute to…

    The objective of this project was to implement and explore various components of a deep learning pipeline for semantic segmentation. Semantic segmentation involves classifying each pixel in an image into specific categories, enabling precise identification and localization of structures or abnormalities. The project addresses a critical aspect of medical image analysis, where precise segmentation is pivotal for accurate diagnosis and treatment planning. The implemented techniques contribute to advancements in deep learning applications for healthcare, showcasing the potential of semantic segmentation in medical imaging.

  • Sparsely Learned Random Walker for Seeded Image Segmentation

    -

    Implemented Sparse Random Walker in a U-Net for seeded image segmentation with limited annotations. Evaluated its adaptability through transductive and inductive experiments, showcasing robust performance in practical scenarios.

  • Depth Estimation in Images using Stereo Vision

    -

    This project centers on the implementation and evaluation of various techniques like window-based and Viterbi scanline-based stereo computing disparity maps from stereo image pairs. The primary objective is to enhance depth perception in computer vision applications. The stereo vision system calculates the disparity map, which represents the pixel-wise differences in the images, allowing the estimation of the relative depth of objects in the scene. The project also investigates the influence of…

    This project centers on the implementation and evaluation of various techniques like window-based and Viterbi scanline-based stereo computing disparity maps from stereo image pairs. The primary objective is to enhance depth perception in computer vision applications. The stereo vision system calculates the disparity map, which represents the pixel-wise differences in the images, allowing the estimation of the relative depth of objects in the scene. The project also investigates the influence of parameters such as window size and regularization weights on the quality of disparity maps. Additionally, it explores the optimization of locally adaptive regularization weights based on intensity contrast. This adaptive approach aims to align disparity boundaries with image edges, enhancing the overall accuracy of depth perception.

  • Semi-supervised Low-level Segmentation using Graph Cuts

    -

    This project focuses on implementing an interactive seed-based segmentation using s/t graph cut for semi-supervised low-level segmentation. The task involves developing a graph cut algorithm that minimizes the loss, combining seed-based constraints and basic pairwise regularization.

  • Image generation for MNIST using VAE and GANs

    -

    Implemented a variational auto-encoder (VAE) and a generative adversarial network (GAN) in PyTorch to generate images similar to those in the MNIST dataset

  • Variational Autoencoder Modification using Spiking Neural Networks

    -

    Introduced modification to VAE by incorporating SNNs with GLIF neurons, enabling accurate representation of complex neuronal processes. Trained and evaluated models on CelebA dataset, generating high-quality images with extended training epochs and longer spike-trains. Provided diverse and representative latent space, surpassing ANNs despite increased training time.

  • Seq2Seq Translation using RNN and Transformers

    -

    The project aims to build a seq2seq model to translate from French to English and vice-versa using RNNs with and without attention and Transformers

  • Island Genetic Algorithm in the context of Vehicle Routing Problems

    -

    Implemented a Parallel Genetic Algorithm for solving Vehicle Routing Problems (VRPs), utilizing the island model approach that divides the population of solutions into separate islands that evolve independently. The GA was able to consistently find high-quality solutions to the VRP with a monotonic decrease in path cost, converging to the local minima

  • Appliance energy Prediction using Prophet and Polynomial Regression

    -

    Forecasts energy usage of households by analysing past data by using FB Prophet and Polynomial regression, and presenting a comparative study of both.

    See project

  • Image Encryption using Enhanced Fiestel Network Algorithm

    -

    The project aims to create an encryption scheme, by combining block shuffling and then using blowfish algorithm to encrypt an image to prevent unauthorised access.

  • Emotion Recognition using Eigen Faces

    -

    Uses eigen faces to predict the facial expression of the person in a live feed and swap their face with the appropriate emoticon. Achieved an overall accuracy of 86% in classification through the Fisher Face classifier by utilizing HAAR cascade for face detection and eigenfaces for feature extraction.

  • Heart Disease Prediction using Machine Learning

    -

    Uses machine learning algorithms like decision tree,Naive Bayes, KNN to predict whether a person is suffering from a heart disease or not.

    See project

  • RANK ORDER LIST OF HOSPITALS FOR MEDICAL GRADUATES

    -

    It ranks hospitals in the US according to the criteria provided by the user,
    according to specialties of their choice.

  • Face Recognition for Home Security

    -

    Uses HAAR Cascade for distinguishing between known and unknown faces

  • Spatiotemporal Oxygen Production Estimation Based on Land Vegetation Using Image Processing

    -

    The project aims to estimate the forest cover loss and also oxygen production over a period of 10-20 years of a particular area using satellite images.It also identifies the leaves from these areas and uses them to identify the plant species, chlorophyll and nitrogen content in the leaves using feature extraction and Support Vector Machines (SVM).

  • SOFTWARE FOR GST INVOICE ENTRY

    -

    The project stores and retrieves GST invoices for different merchants using Java, JSP, Servlets and HTML.

More activity by Sonal

View Sonal’s full profile

  • See who you know in common
  • Get introduced
  • Contact Sonal directly
Join to view full profile

Other similar profiles

Explore more posts

Explore collaborative articles

We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.

Explore More

Others named Sonal Bihani

6 others named Sonal Bihani are on LinkedIn

See others named Sonal Bihani

Add new skills with these courses

See all courses