Fruit Grading and Sorting through Image Processing

This project is a part of my Bachelor's thesis. It involved the implementation of some image processing and machine learning algorithms based on a research paper to sort tomatoes on the basis of their physical attributes, which is a great initiative for the fruit industries considering the growing needs of quality. Currently, I am working on the same project to implement some more related algorithms and compare their performances. I plan on publishing the paper to some journal after I finish with the research work.

A 3x3 FPGA Overlay Architecture

A course project in Master's degree program that involved the use of VHSIC Hardware Description Language (VHDL) to implement an FPGA overlay architecture. This 3x3 FPGA had 9 Configurable Logic Blocks (CLBs) and a full adder was mapped on 2 of the logic blocks. Generally, it could map any logic on its K-LUTs; where K could be 1-4.

A Statistical model for prediction of ODI stats of a cricket player

A course project in Master's degree program that implements some statistical methods on 3/4th of the given ODI data of a cricket player to predict the runs, centuries, ducks and other stats for the remaining 1/4th of the data. The data for this project was obtained from some online websites that provide detailed cricket statistics. The project hasn't yet completed, but upon completion, I plan on publishing a paper on it.

Implementation of some Networking concepts in NS-3

A course project in Master's degree program that involved the use of most common network simulation software, NS-3 on a Linux distribution to simulate transmission of packets over the network. The first part of the project transmitted packets from one node to another using a routing node in the network. The second part involved the implementation of a packet transmission system in which random Poisson data traffic was generated using OnOffApplication specified in the NS-3. This traffic was then used to simulate queue paramters using a Tracer file.

Network Traffic Classification using Machine Learning and Weka

Traffic classification is of fundamental importance to model the network behavior. A classified traffic information is essential to understand how resources in the Internet are being used to effectively control the services that receive this traffic. The traffic classification and identification is vital to the areas of network management, network planning and QoS provision. This course project involved a traffic classification scheme based on machine learning (ML) algorithms. A ML software, WEKA (Waikato Environment for Knowledge Analysis), is utilized to perform this experimentation which contains all the implementations of these algorithms.

Image Compression using Haar Wavelet Transform and Run-Length Encoding

The Wavelets are a powerful set of tools to deal with different scientific and engineering problems, including, but not limited to the signal compression or de-noising, image enhancement, compression or de-noising. This implementation focuses on the compression of images using the famous Haar Wavelet Transform, which is relatively simpler of all the other 2-D Wavelet transforms. The selected image is also quantized and Run-Lenth Encoding (RLE), which is a lossless data compression technique, is also applied on the image.

Wavelet Feature Selection for Image Classification

Feature selection is the process of automatically selecting a subset of most relevant attributes in the given data. A real life data may contain a large number of features or variables that are redundant or irrelevant. Feature selection technique is used to efficiently extract only the useful and relevant set of features from this huge data. There are different transformation methods such as Independent component analysis, Principal Component Analysis etc. Wavelet feature extraction is also one of the techniques used to extract the most relevant features in a set of data. This course project involved the re-implementation of a paper that uses Wavelet Packet Transform to perform the feature extraction using two novel algorithms, Mutual information based sub band selection (MISS) and Sub band Grouping and Selection (SGS).

Descriptive Statistics Final Project

This is the final Project for the Descriptive Statistics course at Udacity. This project is done by conducting an experiment dealing with drawing from a deck of playing cards and then creating a writeup containing my findings. A Matlab code is written to perform this experiment.

• Code

Color based Object Tracking

The project focuses on implementation of the object detection and tracking system based on its color. The input to the project is the video/image data which is continuously captured with the help of a webcam interfaced to the laptop. The high level vision control system uses serial communication to direct the low level mechanical parts. So, if the object is detected, the servo motor attached to the camera is rotated in such a way that wherever the object moves, the camera points to that object.

Temperature Logger

Monitors the temperature of temperature-sensitive pharmaceutical products in a fixed storage area and uploads the logs to an offline/online server.
(Arduino, C/C++, Python, PHP, MySQL, Javascript, jQuery)

Capture

Saves the time by extracting motion frames in long videos in few seconds.
(RaspberryPI, Python)

Volumetrics

Finds the Volumetric weight of a box through Image processing.
(Arduino, RaspberryPI, C/C++, Python)

Person Detector

Triggers an alarm if a human is detected in a real-time video.
(RaspberryPI, Python, C/C++)