An Image Processing and Machine Learning Desktop Application coded in Python. The System would take in a leaf image as an input and provide the predicted name of the tree the leaf may belong to as an output. SVM, KNN and MLP models were used to predict the output.

Where all the current state of the art deep learning models failed to segment highways with satisfactory accuracy due to imbalanced and high confusing classes like roads and canals included in the training dataset. This Capsule network was able to provide near perfect results.

This SegNet model has no need for colored imagery as it is able to provide accurate results of segmenting orchards from images containing forests and fields by learning solely from the symmetric textures created by the trees planted in the orchards.

This deep learning model uses a proposed method which is able to enhance latent features in an image, thus allowing the encoder to properly distinguish building terrace from its side wall in an angled satellite imagery. After the terrace is extracted an algorithm is used to convert it into composite rectangles for footprint calculation.

As the name suggests, the application recognizes common ASL words in a video. It uses Tensorflow PoseNet to extract the position of key body joints at different time steps, these features are then used to predict the output.

A comparison of two custom non-uniform blind deblurring deep learning networks for sharpening images of general dynamic scenes irrespective of their resolution and noise combination. The models included a Capsule Network and a multi-scale Convolutional Neural Network.

An implementation of an automation system for the transportation of items in a distribution center. The system uses a declarative programming language, [CLINGO] to output a minimum step action plan for robots that completes the given list of orders while satisfying all the constraints.

A Web Application based on the [VASSL] paper, made better using Data Visualization and AI techniques to provide a better and efficient process in identifying spam bots by looking at their tweets, comments and other account activity at different aggregate levels.

A file-sharing, and storage website written in PHP where users can browse, upload and download various resources irrespective of file format with multiple levels of file-privacy.

The dashboard is a web based professional trading view of multiple charts to track stocks. It is designed using visualization tricks that help novice traders understand similarity between stock trends and thus give them the ability to make better and safer decisions.

Applied a novel approach to focus on specific context text via question hypothesis with few architectural changes in a multiway BERT attention model and got a 15% better accuracy on [CosmosQA] dataset than the baseline BERT model.

A Machine Learning, NLP and Analytics project that deep dives into the various features a question possess to help predict its quality. The objective and subjective features extracted were able to summarize most of the textual, structural, readability, and semantic details contained in a sample.

A four rotor drone built using Arduino and controlled via mobile application on which a live feed from the drone could be seen. The drone used its on-board gps coordinates and the phone's coordinates to follow the phone wherever it goes.

An arduino based, remote controlled battle bot designed with a spiked, spinning drum at its front as a weapon.

The system consists of a Microphone, Speaker, and a Laser. The laser beam is used as carrier, which is modulated by the signal to be transmitted. On the receiver end, the desired signal is separated from the carrier. A wireless laser link is thus used to transmit information from one end to another in its line of sight.

A Wireless Android operated two motor Boat built using Arduino with Bluetooth connectivity.

A four wheeled, wired, remote controlled, car that also uses information from various sensorslike proximity sensor, light sensor and temperature sensor to self control itself.