Projects

AI-Project-For-Healthcare for Covid-2019

This project is included artificial intelligence, machine learning, data science, computer vision projects related to healthcare.

Breast Cancer Classification: The aim of this project is classifation the tumors into malignant or benign with machine learning techniques.
DNA Classification Project: The aim of this project is to find out whether the DNA sequence is the promoter.
Detecting COVID-19 with Chest X-ray using PyTorch:
Diabetes Prediction with PySpark MLLIB:
Heart Failure Data Analysis:
Relationship between COVID-17 & Happiness in that Country:

Organized the database into 3 folders (train, test, val) and contains subfolders of each image category (Pneumonia / Normal).
Contains 5,863 X-Ray images (JPEG) and 2 categories(Pneumonia/Normal)
Used chest X-ray images (anterior-posterior) which were selected from retrospective cohorts of pediatric patients of one to five years old from Guangzhou Women and Children’s Medical Center, Guangzhou.
Designed CNN models for training to optimize weights using PyTorch.

With a team of 3 AI engineers in building a multi-object-tracker using Python Deep Learning Framework - Tensorflow.
Developed project base on multiple-object-detecting project which was developed by C++ & OpenCV.
Used EMGU-CV with C# for preprocessing and implemented autoencoder to get high accuracy. Designed several Tensorflow Models for Object Tracking and trained for getting best optimized weights (coefficient values) and chose the best models and implemented forward processing project.
Created CI/CD pipelines using Jenkins for developing project, modulated the final project and published so that it can be used for further project relating object .

Put simply, our full-frame binary detection (FireNet, InceptionV1-OnFire, InceptionV3-OnFire, InceptionV4-OnFire) architectures determine whether an image frame contains fire globally, whereas the superpixel based approaches (SP-InceptionV1-OnFire, SP-InceptionV3-OnFire, SP-InceptionV4-OnFire) breaks down the frame into segments and performs classification on each superpixel segment to provide in-frame localization.
For the best detection performance, used InceptionV4-OnFire which operates at 12 fps, for best throughtput (17 fps) use FireNet which has slightly lesser performance.
Used pre-trained network models from http://dx.doi.org/10.15128/r19880vq98m

FireNet architecture

Interception V1-OnFire architecture

Interception V4-OnFire architecture

Interception V3-OnFire architecture

Original frame (left), Frame after superpixel segmentation (middle), Frame after superpixel fire prediction (right)

This Project is the Tensorflow implementation of Convolutional Pose Machines, one of the state-of-the-art models for 2D body and hand pose estimation.
Used OpenCV and Tensorflow, and Created dataset for training.
Designed own Tensorflow model for training and implemented a distillation training scheme.

This project is designed for streaming speech recognition