DoubleMap on Alexa

I love working on real problems that does makes some difference in the real world. Started as a personal project duing summer! Tried modeling DoubleMap application to Alexa using flask-ask. Using this Alexa app, we can get the current location, time of the bus that will arive at a given stop and so on.

• Date: June 2018
• #Python #Flask-Ask #JSON

Close Project

Image Orientation Classifier

A neural network used as a classifier that decides the correct orientation of a given image. Dataset used was Images from the Flickr photo sharing website, 10,000 Training set and 1,000 Test set of Images. Each raw image is converted into numerical feature vectors, i.e., n × m × 3 color image (the third dimension is because color images are stored as three separate planes of the red, green, and blue), and append all of the rows together to produce a single vector of size 1 × 3mn.

• Date: November 2017
• #Python

Close Project

Part of Speech Tagging

Created a tagger to mark part of speech of every word in a sentence from the text corpus of 1 million words using the implementation of Viterbi Algorithm on hidden Markovs model, as a NLP task.

• Date: November 2017
• #Python

Close Project

Shortest Path Finder

Developed a model using 4 algorithms simultaneously (A* search, DFS, BFS, Uniform Cost Search) to find the shortest distance between the data set of 12000 cities of United States.

• Date: September 2017
• #Python

Close Project

Image Warping

This project had three parts. Every part is connected to form the holistic approach of warping the image and projecting to different coordinates of another image.

1) The first part contains warping of the image. To get a perfect warping, we applied the 3*3 transformation matrix using homogenous coordinates. We have used inverse warping technique to warp the images for the new transformation matrix.

2) This image extends the 1st part as in this part, we find 8 unknown points, create a transformation matrix, and then perform the warping same as we performed above using inverse warping.

3) This part will project the image to the different billboards. We used the coordinate of the image that is to be projected on to the billboard and the coordinate of the billboard on which the image has to be projected. This is the combination of both the steps that are performed above. First, we find the coordinate of the image and billboard and then use the same matrix to find the 8 unknown coordinates and then using this transformation matrix, using inverse warping, we projected the image onto the billboard.

• Date: March 2018
• #C++ #cimg

Close Project

Algorithms and Data Structures

This was not exactly a project but it has interested and grabbed my interest throught out. I like to implement different algorithm and visualize how beautifully and structrally they are created. As in my fun time, I like to code different data structres and algorithms and analize there efficiency and potency.

• Date: Forever :)
• #Python

Close Project