FisMat

Inspiration

Our main inspiration came from the use of new age model optimization technologies like OpenVino, as well as FRIDA, Softtek’s GenAI. We felt like these two technologies could be big helpers of a general community of people who didn’t have the access to higher-end processors and general hardware. To make sure we fulfilled these goals, while having a business focus, we wanted to create a model that could financially help small businesses.

What it does

The model allows for the business to upload a picture of their pending bills, our model extracts the text from the bill, to then be sent to FRIDA AI, who will effectively act as a financial advisor, generating a financial advice report, giving insights of financial health and actionable recommendations, thus improving the financial management of any small business

How we built it

We started off by optimizing the Tesseeract-OCR, using OpenVino, and we managed to create a model that recognized words and numbers in an image. However, further programming drew us to the conclusion that the model before being optimized would be of better functionality for us. This way, using pytesseract, javascript and CSS, we were able to localhost a model page for our model, that includes the use of FRIDA, this way, the only thing the user has to do, is upload their bill image, and they receive a full financial report.

Challenges we ran into

The challenges we had to face throughout the hackathon included settling on an idea. The fact that the challenge was open-ended, allowing us to address any problem, made it difficult to act quickly. Additionally, we were stuck for a few hours due to mental and physical exhaustion. We also had to deal with the use of new libraries, the terminal, and using Git for the first time. Without any doubt, it was challenging to get comfortable with those technologies, but we managed to adapt in the best way possible and were able to complete a project that left us with many valuable lessons.

Accomplishments that we are proud of

We are very proud of our final model considering this was our collective first hack, we feel very happy to have been able to develop this model and this idea, while learning to use new technologies like OpenVino and FRIDA. We are also very proud of the fact that we managed to create a solution that can help businesses of our community.

What we learned

We learn a lot of new technologies such as FastApi, Frida, and OpenVino. It was a day full of surprises and programming errors. Thanks to that, we developed many competences in AI programming, general python programming, usage of GitHub and web development. All these topics were really interesting and challenging to comprehend but in the end we were able to use them in our favor and complete our project.

What's next for FisMat

For FisMat, the next steps involve continuing our learning journey, delving deeper into the fields of AI, Machine Learning, and Deep Learning. This project has provided us with the opportunity to study these topics more thoroughly and effectively. We aim to keep honing our skills through practice and participate in various events to further enrich our knowledge. Our ultimate goal is to build a future in this area that has ignited a deep passion within us.