Abstract - The task of sign language translation is crucial for enhancing communication for the deaf and hard-of-hearing community. In this paper, we build upon previous work by enhancing the large language model (LLM) component of a hybrid approach for sign language translation from continuous video streams. The prior hybrid approach employed a GPT-3.5 Turbo prompt to generate translations from identified sign spottings. We improve this by incorporating fine-tuning techniques and revising the prompt. We employ a sign spotter from existing literature to identify individual gestures within the video stream. These identified gestures are then processed by an LLM, which constructs grammatically correct and coherent sentences. Our evaluation of two models demonstrates significant improvements, with the fine-tuned Gemini-1.0 Pro model showing the most successful enhancement in translation accuracy and coherence.

Abstract - Currently, LiDAR has become ubiquitous for autonomous vehicles due to its distinct ability to capture dynamic changes in an incredibly detailed map of the vehicle's environment in real-time. In comparison to other sensors (such as stereo camera vision), LiDAR is versatile to long distances while still maintaining high definition. However, deploying and testing these systems presents a large caveat to any motivated entity who is looking to do so. Logistically, testing these models and training them on physical vehicles requires extreme resources and are computationally expensive. Though, through the use of a simulated environment testing is much more feasible. So, in this regard, I methodically approached a solution to LiDAR odometry to localize a vehicle in real-time, using sensor data to create a trajectory path. Subsequently, I used the CARLA simulation environment to test varying vehicle environments and sensor conditions.

I watched Interstellar and decided to build an autonomous tractor company. I built a team of nine guys from MIT, Vanderbilt, CMU... and we built it in 30 days.

Think Duolingo and a typing game, but with sign language. The web-app uses a device's built-in camera and pulls data in real-time to detect hands, facial expressions, and body position to determine what gesture the user is making. Our web-app dynamically looks for temporal motion and decides whether or not one is making the correct sign. This isn't any boring language learning app, it's a game. A user is able to keep track of their signing speed and accuracy in the app (much like a typing game).

Since the pandemic, I've found a lot of joy rebuilding and selling computers in bulk to various businesses, religious institutions, and local buyers. However, a lot of my sales have been expanding internationally where a bulk of the computers end up in school districts in African countries.

If you're not familiar with stenography and stenographic machines, the biggest takeaway you should know is their cost-to-entry (thousands). Seriously, people go to college just to learn how to type on these things. After learning about this I decided to try and develop a steno keyboard of my own. I developed one for less than $15 with my own circuitboard and open-source code, and then began a kickstarter campaign that sadly failed due to other (objectively better) competition.