Research project in graduate school focusing on non-audible speech patterns. Presented at the 2019 International Conference on Computational Science and Computational Intelligence (CSCI), Las Vegas, NV.

Abstract

Research advancement of human-computer interaction (HCI) has recently been made to help post-stroke victims dealing with physiological problems such as speech impediments due to aphasia. This paper investigates different deep learning approaches used for non-audible speech recognition using electromyography (EMG) signals with a novel approach employing continuous wavelet transforms (CWT) and convolutional neural networks (CNNs). To compare its performance with other popular deep learning approaches, we collected facial surface EMG bio-signals from subjects with binary and multi-class labels, trained and tested four models, including a long-short term memory(LSTM) model,a bi-directional LSTM model, a 1-D CNN model, and our proposed CWT-CNN model. Experimental results show that our proposed approach performs better than the LSTM models, but is less efficient than the 1-D CNN model on our collected dataset. In comparison with previous research, we gained insights on how to improve the performance of the model for binary and multi-class silent speech recognition.

Link to Paper

Link to Article

Proof of concept project developed with other members of a team. The project focused on creating a marketable IoT device using lean start-up methodologies.

Information

The ease fall detection system uses sound, infrared, and vibration sensors to detect when a fall has occurred, and promptly notifies designated users if their loved ones have fallen.

The user experience lies in the mobile app. Once a fall has been detected by the ease sensor system, a user is notified on their mobile phone. Inside of the mobile app, the user has the option of calling the home where the fall took place or calling 911 directly.

Arduino based IoT alerting system that lets users know when their bike locks are being tampered with by sending SMS messages.

Abstract

Bike theft is an increasing problem that is relatively low priority in the eyes of the police due to their low value. Current deterrents, such as bike locks, are no longer considered to be secured due to being easily defeated. This paper is used to inform its readers of the development of a system used to deter and counteract against bike thieves. This is achieved by using an Arduino Uno along with a vibration to detect any would be thieves. A piezo buzzer sounds an alarm if any vibration is detected in order to scare away any thieves and alert any nearby individuals. A GSM module is also integrated in order to send a text to the owner of the bicycle of a possibly hijacking. This system is armed using a 4x4 keypad along with a red and a green LED as an indicator to display the status of the system.

Using general web design skills such as HTML, CSS, JavaScript, and Bootstrap to build websites and IoT based applications. Built using backend frameworks such as Django and Flask.

theflowerdoc.com

Django based website built using the Bootstrap framework and PostgreSQL backend database. The website was designed for the client to showcase their work, add personal blogs, and have customers request any services. An admin page allows the client to personalize their own content.

TheFlowerDoc.com

rommelfernandes.com

Personal website designed using Flask, Bootstrap, Jinja, JavaScript and a PostgreSQL backend database. Managed through Github actions CI/CD. Designed to relate project information and handle Internet of Things API requests.

RommelFernandes.com

Using transfer learning to build a model for an existing hot wheels collection. Model will be deployed using REST API and utilized in a mobile application.

More to come...

Taking a stock Power Wheels vehicle and adding microcontrollers to automate driving. Autonomous driving model will be trained using TensorFlow and deployed on microcontroller using TensorFlow Lite and quantization. Microcontroller will then have the ability to control speed and direction of vehicle using streaming video from a camera attached in the front.

More to come...