Santhosh Kumar's Projects
Repo contains Sentiment analysis model built on Amazon Food Reviews for Sentiment analysis.
Repo has PyTorch implementation "Attention is All you Need - Transformers" paper for Machine Translation from French queries to English.
Blind Deblurring (Deconvolution) System
When Processing Documets of like forms and statements, information present could be of various types. Checkbox items, One such type, are found commonly in the forms/douments. identifying checkboxes and whether a check box is selected or not forms vital part of information extraction. This project outlines a simple approch for processing checkboxes and extracting information from them. The idea of the project is let users use this as ready to go for similar usecase without or with little modifications.
Various types of information can be present when processing documents like forms and statements. The checkbox items, one type of item, are found in many forms and documents. A vital part of information extraction is identifying checkboxes and determining whether they are selected or not.Β An approach is presented for extracting information from checkboxes in this project. This project is designed to be used as a ready-to-go solution for similar use cases with minimal modifications.
Simple tool to split COCO annotations into train/test datasets.
The 3rd edition of course.fast.ai
Simple transformer implementation from scratch in pytorch.
testing pull and push
Implemented image caption generation method propossed in Show, Attend, and Tell paper using the Fastai framework to describe the content of images. Achieved 24 BLEU score for Beam search size of 5. Designed a Web application for model deployment using the Flask framework.
Source code of Indian Paper Currency Prediction :fire:
(DEPRECATED) Export public LinkedIn profile to Markdown format
The repo contains the setup files for developing my blog website using Django framework
The repository is implementation of Netflix recommender prize winner's solution
This project is about building Deep Learning CNN model to label satellite images with Atmospheric conditions (Multi-label classification problem)
This project involves application of PCA technique on image data and assessing its performance in terms of information retention and compressibility.
Project is on building an ML model along with text feature extraction to identify similar questions posted on Quora.
Data science portfolio