ð Garde: AI/ML Fencing Coach built using a bleeding-edge web dev tech and AI/ML tech stack ðĪš
ð About Garde
Garde is an automated fencing coach being created alongside Davis Fencing Academy, a Team USA Fencing certified association. Garde aims to perform a robust analysis of a fencer's form and harness generative A.I. to give detailled feedback as to how the fencer can gradually improve themselves over time. In addition to this feature, Garde has recently implemented a feature which allows fencing coaches to access their students' videos and give them individual feedback as well as help them create fencing routines. We hope that Garde is used by leading fencing institutions not only in Davis and The Bay Area, but also worldwide as this program is truly one of a kind and is currently in the midst of being patented.
ðĨ Tech Stack
Currently, Garde uses Next.JS and Tailwind CSS to create an intuitive and immersive frontend that clients can easily learn and adapt to. We are also planning to incorporate Framer Motion and other animation libraries to enhance user comprehension of Garde. Clerk is being used for user authentication as it's easily scalable and offers a seamless transition between the landing page and the pages which the fencer and coach can access. For the backend, Garde uses Amazon S3, MongoDB, Prisma, Tensorflow.JS, Plotly AND OpenAI API to store and fetch valuable user data, use ML to analyze and visualize fencer form, and use generative AI to give valuable feedback that the fencer can use.
ð· Screenshots
The landing page for Garde, it uses Spline to render the 3d spheres. This page gives users all of the information they need about what Garde does before they decide to use it. Also, this page uses EmailJS so that coaches who are interested in using the product can send us an email so that we can give them access
The fencer page for Garde, this page is used by fencers to perform the instructions shown above the canvas and get the feedback they need to improve. Garde calculates the angles between their joints, speed and distance in real-time to offer them key information they can use to assess their performance.
The coach page for Garde, this page allows the coach to see all of the videos of their students and give them feedback on the right. The textbox is made using TipTap Editor and offers easy customizability. Once the coaches are done seeing the students' performance, they can also view their progression over time which is detailled through Pi charts and line graphs.
ðĨïļ Code
Currently, Garde is patent pending and a lot of the code is private. However, if you are curious about the code for Garde and would like to have a sneak peek. Please contact Aggie Sports Analytics via email to schedule an appointment.