Frontend Overview
HomePage UI
Features Section
Video AI-Analysis UI
Evidence Scraper
Testimonials (temp.)

Inspiration

As members of the high school speech and debate team, we recognize the importance of having a coach to guide us in honing our debating skills. Currently lacking a dedicated coach, we feel the need for personalized assistance to elevate our craft. Additionally, being a small team poses challenges in swiftly sourcing relevant evidence during debates. Recognizing these obstacles, we were motivated to find a solution that not only addresses our immediate needs but also contributes to the broader improvement of our debating abilities.

Problems we took into account as "Inspiration": • Inaccessibility to debate tournaments • Underperformance in speaking abilities upon various groups • Limited access to affordable debate tutoring & tournament participation --> barriers to individuals facing economic challenges

What it does

ArguMentor addresses ALL of these pain points, as it has an AI speech coach to track stuttering and eye contact that the speaker needs to maintain throughout the speech. It can give a score based on your performance and give some feedback so that you can take the correct course of action.

MAIN FEATURES:

Eye Contact Detection • The app can use machine learning and computer vision to track the percentage of time that you have been using eye contact, and then provide you feedback on how to improve, as well as other key metrics.
Stuttering Detection • The app can find out how many times you have stuttered so that you can avoid it during your speech. It displays a count so that you can take the correct course of action.
AI-Optimized Evidence •Combining the powers of both generative AI and Google, the app can pinpoint pieces of evidence from the internet and summarize them effectively. This is especially useful for finding evidence quickly.

How we built it

As for our tech stack, we used React, Bootstrap CSS, and Vanilla JS for the front end. For the backend, we utilized the NumPy, Tensorflow, and OpenCV python libraries for ML, and LangChain for the generative AI portion of the evidence finder.

We have attached code snippets within our video. If you'd like to check out our full code, you can do so by checking our Github repo.

Challenges we ran into

Flask Routing Complexity -- Initial struggles with understanding and implementing Flask routing. • Required concerted effort to identify and resolve issues.
Problem-Solving Journey • Perseverance was crucial in overcoming challenges. • Engaged in a systematic problem-solving process.
Knowledge Enhancement • Successfully navigating through challenges contributed to a deeper understanding of the Flask framework.

Accomplishments that we're proud of

We were proud that we could train the voice and audio model, even without any references online. We were also proud of the front-end design that we were able to create. Finally, we were also proud of the accuracy of the evidence finder model.

What we learned

Insights Gained
Efficient Routing and ML Integration	Acquired skills in effectively routing and seamlessly connecting our machine learning components to the front end. Gained practical knowledge in optimizing the flow of information between different layers of our application.
Frontend Model Creation	Mastered the art of creating frontend models, enhancing our understanding of user interface design and implementation. Developed proficiency in translating design concepts into functional and visually appealing front-end components.

These newfound skills contribute not only to the success of our current project but also serve as valuable assets in future endeavors.

What's next for ArguMentor

The next phase for ArguMentor involves creating an ArguMentor chatbot. We're exploring social media marketing, with platforms like TikTok and Instagram to broaden our reach and engage with a wider audience. This will allow us to further expand our initiative to individuals who are interested in genuinely utilizing our application for the better. Additionally, we're developing a leaderboard system based on usage, utilizing Arg Points as a rewarding mechanism.