<h3>Arieal Cactus Identification</h3>

Arieal Cactus Identification

This was kaggle competition where I have to create Machine learning model that identify whether there is cactus in an image or not. Image is taken from Ariel view. Here I tried 2 different version. In first version I tried to use transfer learning where I used pretrained model weights as base weights and train them on cactus idenfication. Model I used for transfer learning is Inception Model. Second method was creating my own Convolution Neural Network (CNN) and train on the images.I got the accuracy of 99%.

PythonTensorflowClassification
<h3> Exploratory Data Analysis </h3>

Exploratory Data Analysis

This project was gettin hands dirty on Datasets where I pick data from websites like Kaggle and UCI Dataset and do some data processing and visualization. Keep in mind no ML model's were created just data preprocessing and Visualization. Most used library was Sklearn and Pandas to read csv files.

PythonSklearnTensorflow
<h3>Advance Regression Techniques</h3>

Advance Regression Techniques

In this project I try to use advance regression techniques like Generalization Stacking(Blending) and weighted vote regression. Multiple models were blended together to predict the house prices. Regression models used in this project are Elastic, Lasso, Ridge, GradientBoostingRegressor, RandomForestRegressor, SupportVectorRegressor, XtremeGradientBoosting(XGB) and LightGradientBoosting(LGBM).

PythonSklearnXGBoostLightGBM
<h3> Diabetes Detection using k-NN </h3>

Diabetes Detection using k-NN

This project was my first time I learned about k-Nearest Neighbors. I thought this was such a powerful algorthim for classification and tried it on diabetes patient dataset.

PythonSklearnOptimizing HyperParameters
<h3>Classifying Fashion Apparels</h3>

Classifying Fashion Apparels

Many people start their first CNN project as MNIST ie. Digit recognition I wanted something different so I used Fashion MNIST 😂. In this project I created Convolution Neural Network(CNN) in Keras to classify different fashion apparels. In the end of the notebook I try to visualize what CNN learned from the dataset and we can easily get insight how CNN learn different parameters on different layers.

PythonTensorflowCNNKeras
<h3> Road Lane Detection</h3>

Road Lane Detection

This is an ongoing project where I use Python to detect road lane. I have used simple approaches for lane mark detection. I wanted to make this the part of self driving project which I will start later this year.

PythonOpenCVImageProcessingONGOING