Mahesh Deshwal's Projects
End to end Named entity recognition from EDA (what it means and reflects) to Explainable predictions (how to properly understand results) and ideas on improvements
Bing Chat automated with Selenium Web Driver to exchange messages
Links to Colab notebooks from my blog, if you have found them somehow or my random experiments
Object Detection, Object Tracking, Object Counting and more using old school CV algos, OpenCv algos and more advanced YOLOv4, YOLOv3
Unofficial modification for the Paper: CURL: Neural Curve Layers for Global Image Enhancement (ICPR 2020) which supports deploy ready code in CPU and GPU both
Complete and Thorough data visualisation tutorial on using Matplotlib and Seaborn from beginning to advance level with in depth customisation and demo of multi variable plots such as Pairplot, Relplot and Catplot
Just some stuff for Interview questions, books, annotated paper, notes, cheat sheets etc etc related to ML,AI, Deep Learning and Data Science
Different methods to clean tweets based on requirements
This is the repo for our new project Highly Accurate Dichotomous Image Segmentation
custom DS and Algo implementation with advance Python based functionalities using dunder methods
Detect the face using EigenFace, SVD and PCA. Implementation of LDA, NaiveBayes Classifier from scratch
App using Python FastAPI + Mongo + Docker + Elastic + Logstash + Grafana + Prometheus + github actions + Kubernetes (K8)
Inspired from the original paper by Stanford's NLP team
3 Part Series to know EVERYTHING about Image Processing using sk-image, PIL and OpenCV
Convolutional Neural Networks to predict the aesthetic and technical quality of images.
Building an ID3 Classifier from scratch and integrating with the sklearn to produce cross validation results. Comparison of DecisionTreeClassifier is also shown
Deep Learning and Keras tutorials for very beginners using Bengali.ai Grapheme prdictions compitions with 0.933 public score
Keras implementations of Generative Adversarial Networks.
Tutorial of how you can build a Custom Layer, Custom Model and a Custom model which has Custom Layers in it
A bot that scrapes your jobs in real time, sort them according to preferences and runs an alert
Unofficial Modification which supports deploy able code for the paper "High-Resolution Photorealistic Image Translation in Real-Time: A Laplacian Pyramid Translation Network"
Starting with Helper Functions
All Helper Functions Ever Needed to build a ML model. Wrapper implementations of EDA, Image Processing, NLP data cleaning and so much more......
Repo for ML Models built from scratch such as ncludes Linear, Logistic Regression, PCA, LDA. CNN, LSTM, Neural Networks using Numpy only
Multiple NER approaches spacy, Keras Dense Layer, Keras CRF layer, Glove , FastText Embeddings , Attention etc
A complete and thorough tutorial on Basic,LSTM (Bidirectional and Stacked) with GloVe embeddings
Scoring an essay using n-gram, Linear Interpolation smoothing and perplexity
National Stock Exchange (NSE), India based Stock screener program. Supports Live Data, Swing / Momentum Trading, Intraday Trading, Connect to online brokers as Zerodha Kite, Risk Management, Emotion Control, Screening, Strategies, Backtesting, Automatic Stock Downloading after closing, live free day trading data and much more
Image Deskewing, Projection Profile, Binarization using thresholding, Noise removal and Improvement using UNet. Data creation by adding adding Gauss, Salt-Pepper etc Noise, Blurring. Lots more