virajbagal Goto Github PK

albumentations

Fast image augmentation library and an easy-to-use wrapper around other libraries. Documentation: https://albumentations.ai/docs/ Paper about the library: https://www.mdpi.com/2078-2489/11/2/125

aptos-diabetic-retinopathy

My first Deep Learning Competition. Involves classifying images of eyes in one the categories of Diabetic Retinopathy. It is multiclass classification problem.

car-price-prediction-

It was a competition held by 'MachineHack.com'. I finished 8th among more than 1000 participants.

chestxray14-reimplementation

This is an attempt to reproduce the results of the paper: Jointly Learning Convolutional Representations to Compress Radiological Images and Classify Thoracic Diseases in the Compressed Domain, ICVGIP 2018, by Ekagra et al. Process is: 1. Train Stage 1. 2. Train Stage 2. 3. Train stage 3 using weights from stage 1 and stage 2.

commclassroomop

credit-card-fraud-transaction

EDA + TSNE +SMOTE + ENSEMBLING

cyclegan

Simple Implementation of CycleGAN

enterpret_summarizer

This is an assignment by Enterpret for Senior MLE

fmix-paper-implementation

In this repo, I have implemented Fmix, Cutmix and Mixup on the Fashion MNIST dataset using PreAct ResNet18. This is an attempt to reprdouce the results from this paper: https://arxiv.org/abs/2002.12047

folder-structure-conventions

Folder / directory structure options and naming conventions for software projects

getting-started-with-git-and-github

Explaining Git and GitHub.

hypmix-emnlp

Code for HypMix EMNLP 2021 (main)

ieee-fraud

IEEE Fraud Detection was a competition on Kaggle. I was competing in it. This is how I went on to analyse the data. The data was highly imbalanced .

image_cls_app_with_mongo_postgress_minio_mlflow

innoplex-hackathon

Innoplex Hackathon was hosted on Analytics Vidhya website. It was NLP hackathon. I ranked 210th out of 3260 competitors. I have uploaded the photo of the final leaderboard.

keyword-spotting

In this project, we do an ablation study of existing audio processing models for 'keyword spotting' using the Speech Commands dataset. The dataset has 65,000 1 second long utterances of 30 short words in English done by thousands of people.

lepton-identification-using-cnn

Objective was to use CNN for classifying whether the electron is real or fake. Real electron has W,Z or other gauge bosons as its mother. Fake Electron comes from QCD processes. Images of Calotowers within dR=0.4 of electron were used as input to images.

lightweight-gan

Implementation of 'lightweight' GAN, proposed in ICLR 2021, in Pytorch. High resolution image generations that can be trained within a day or two

ml-explainability-and-eda

Heart Disease dataset : EDA, SHAP Values and Partial Dependence Plots

mljar-supervised

Automated Machine Learning Pipeline with Feature Engineering and Hyper-Parameters Tuning :rocket:

mlops-basics

mmbert

MMBERT: Multimodal BERT Pretraining for Improved Medical VQA

multi-object-tracking

In this project, I compare two SOTA Multi-Object Tracking (MOT) algorithms: (a) Yolov5+DeepSORT and (b) FairMOT

object_detection

Implementing algorithms for wheat head detection

polyglotpal

Fun and effective way of learning new languages using AI

product-matching-shopee

This repo contains the code for product matching problem statement. Given image and textual description of different products, the task is to match similar products with each other. Here, I took a semi-supervised approach because the test set apparently contained mostly unseen images and their text descriptions.

pytorch-lightning

The lightweight PyTorch wrapper for high-performance AI research. Scale your models, not the boilerplate.

pytorch-template

PyTorch deep learning projects made easy.

sentiment_app

In the repo, we deploy simple sentiment analysis model to Heroku

textassistant

AI that can quickly summarize and allow Q&A on PDFs, Docs, Images and Youtube videos.