My Projects

Picasso Art Bot - A contour mapping & tracing robot using Raspberry Pi.
Timeline : Jul 2022
  • A hardware implemented robot using Raspberry Pi microprocessor, motor driver and simple robot kit.
  • Developed an interface using python, which will take images that are either stored locally or can be fetched using an external camera.
  • This image now fed to a program that commands a turtlesim bot in a ROS simulation to trace the contour of the input image. The contour can also be tuned with adjustments made to the noise, reducing the complexity etc.
  • Now for every closed loop in the contour, with the help of multi-threading a new turtlesim bot is generated and traces the contour.
  • The tracing is simulated in the interface developed and once the image is traced by all the bots the program now commands the hardware implemented bot to sketch the contour on a piece of canvas.
Technologies Used : Python, C++, ROS, OpenCV.
Pneumonia and COVID-19 detection using Convolutional Neural Network (CNN)
Timeline : Jul 2022
  • A Deep Learning model based on CNN that has the capablity to identify pneumonia and COVID-19 based on the X-ray images.
  • The model is trianed based on Chest X-rays of patients to identify the affected areas of lungs.
      COVID-19
    • the model consists of 3 hidden layers, with "ReLu" as the activation function.
    • The dataset for COVID-19 detection is from Kaggle which consists of nearly 2328 patient's records.
      • Pneumonia
    • This CNN model is consists of 5 hidden layers, with "ReLu" as the activation functions.
    • The dataset for Pneumonia detection is also from Kaggle which consits of two classes : Pneumonia and Normal. There are almost 5863 records in the dataset.
  • Finally using the Pneumonia model we have obtained an accuracy of 98.16% and for the COVID-19 model we have obtained an accuracy of 97.5%.
Technologies Used : Keras, Python, PIL, Matplotlib, Pandas.
Car Price Prediction using Machine Learning
Timeline : Nov 2021
  • Developed a Machine Learning project that uses the car dataset from kaggle, which has features such as car price, kilometers driven, number of owners, fuel type, dealer type etc.
  • This project makes use of Random Forest Regressor for predicting the car price for a given set of featues.
  • The inputs to the model is actually given from a webpage develop using HTML and Flask.
  • Finally when the model is evaluated it acheived an accuracy of 93.4%
Technologies Used : Python, Machine Learning, Flask, HTML.
Sudoku Validator using Multi-Threading.
Timeline : Dec 2021
  • A Multi-Threading based project, the inputs to which are a set of matrcies of 9x9 size (general sudoku size). This model evaluates the given input matrices whether they are solutions of sudoku puzzle or not by checking the conditions for every element in the matrix.

1 Thread model

  • This model makes use of a single thread(instance of an execution of a program) to check for all conditions.
  • This single is used to check for same elements in the column, row and 3x3 grid inside the matrix.
  • The thread is cancelled after every process. This is done for all 81 elements.

3 Thread Model

  • This model create three threads for checking. One for checking in a row, one for checking in a columnn and the other one for checking in the 3x3 row.
  • For every element these 3 threads are created and after checking they are cancelled.

11 Thread Model

  • This uses 11 threads in total, one thread to check every row, one thread to check every column.
  • And finally other 9 threads are used to check for all other 3x3 grids.

27 Thread

  • In this model every row, every column and every 3x3 grid is checked using a seperate thread.

Technologies Used : Keras, Python, PIL, Matplotlib, Pandas.