Projects 📒

Handwritten Character Recognition using Own Dataset

Using SVM and other techniques to create a model which would predict/recognize the english alphabet from our own dataset.

For more Information on this refer my blog : Here

Vehicle Detection and Trip Count Using OpenCV and YOLO

This is my final year project. This basically helps surveillance and helps catch fraud. The intention is to count the number of times a tanker has taken trips to and fro the area of interest.

For further information on the theory and procedure of the project, refer to the blog : Here

We used YoLoV3 to get this job done.

Cassava Leaf Disease Classification Kaggle challenge 2020

Binder

Link to the Blog Post : Here

My Submission Score History :

accuracy

Anime Recommendation Engine using KNN

Using KNN to make an anime recommendation engine based on MyAnimeList Database

Twitter Personality Insights using IBM Cloud Personality Insightsv3

Get the OCEAN personality scores for any twitter celebrity or yourself. This Uses IBM Cloud Personality Insightsv3 Feature and Tweepy to scrape tweets(legally! LoL) using the twitter developer account.

References

nicknochnack
Human rights Implicatiosn of IMB Watson Personality Insights

Kaggle survey 2020 insights challenge

My submission for Kaggle 2020 Survey challenge

Official kaggle competetion link : Here

Fiddle around with the notebook here : Binder

Medical Prescription OCR and medicine/drug details Extraction.

EVALUATE HEALTH

Given any medical prescription (from practo or any such service), extract information about Drug/Medicine, Dosage, Frequency, Form, Duration, Potency, Name of the doctor, name of the hospital, date, age, and other personal patient details. This is an integral part of the product that is being built at EVALUATE HEALTH.

To know how this works exactly go here : Add blog link

Technologies Used : OCR NER Flask DBMS n-grams NLP

Tools Used : Spacy MySQL Flask_mysqldb pandas scikit-learn pytesseract

Test Report and Medical Diagnosis OCR and Data Extraction

EVALUATE HEALTH

Given a medical diagnosis certificate or any test report, extract all the tests' names and its relevant values from it. The file can be PDF, Image or any other relevant file format.This is an integral part of the product that is being built at EVALUATE HEALTH.

To know how this works exactly go here : Add blog link

Technologies Used : OCR Flask DBMS n-grams Fuzzy String Matching NLP

Tools Used : Spacy pandas scikit-learn pytesseract FuzzyWuzzy CosineSimilarity Camelot

Personal Medical Diary For the Patient.

EVALUAE HEALTH

This feature allows the user to maintain a personal medical diary which can be used to give personalised recommendations to the patients based on the medical problem he/she is experiencing. We pick out any diseases, allergies, symptoms or allergies that the user might have and recommend doctors or online services based on them.

To know how this works exactly go here : Add blog link

Technologies Used : DBMS Fuzzy String Matching NLP NER

Tools Used : Spacy MySQL pandas SciSpacy OpenCV

PanCard and Aadhar Card Data Extraction

EVALUATE HEALTH

Simple Data extraction feature that extracts important data from pancard and aadhaar card from the user’s Vault to validate the account. Also added a feature to extract faces of the users and use it as profile picture.

To know how this works exactly go here : Add blog link

Technologies Used : OCR NLP NER

Tools Used : Spacy MySQL pandas OpenCV

Used LSTM on MNIST data

Used LSTM on MNIST Data to classify. Yes, its bizzare.

Non Data Science Related :