The problem SheConfident solves

"We understand that confidence can be a challenge not only for students but also for employees and college students during presentations and important meetings. That's why we created SheConfident, a program designed to help people overcome these obstacles and increase their confidence and self-esteem.

This lack of confidence can hold them back in their careers and prevent them from reaching their full potential. This is particularly evident in public speaking situations, such as presentations and important meetings, where women may feel nervous or unsure of themselves. As a result, they may not effectively communicate their ideas or contribute to discussions, leading to missed opportunities and a lack of recognition for their work.

SheConfident is a program designed to help women overcome these challenges and increase their confidence and self-esteem, enabling them to advocate for themselves and succeed professionally.

Our approach includes using a DL model that helps users improve their voice tone and confidence on camera, as well as sharing their most confident moments with them in the form of a non-fungible token (NFT) that can be listed on our marketplace.

In addition, we offer a decentralized platform for hosting fundraising campaigns, where anyone can list their project and receive support from interested investors and donors. We believe that by fostering a supportive and empowering community, we can help people achieve their goals and grow with confidence.

Challenges we ran into

The problems we ran into while building our solution were:

1. Integration of smart contract with the front end:
   During the initial development of our solution, we implemented the integration of the smart contract with the front end using the ether.js library. However, we encountered issues with different dependencies versions while writing and reading the smart contract. In order to address these issues and improve the security of the smart contract with minimal vulnerabilities.

2. Creating Deep learning model APIs:
   In order to integrate our deep learning model with the front end, we utilized the flask framework to create API endpoints that could process audio and video feeds. The feeds were received in the body of the request and were converted into base64 encoded binary arrays before being processed. Once processed, the results were returned to the client as base64 encoded binary arrays. We encountered issues related to data formatting during this process and implemented the use of encoding and decoding techniques on the backend in order to properly process the data with the deep learning models.

3. Deploying our flask server:
   We attempted to deploy our flask server on various deployment services such as Render, and Railway but these services did not support the use of heavy libraries like TensorFlow and OpenCV on their runtime. As a result, we decided to create a Docker container for the flask server and deploy it on Microsoft Azure App Services. This allowed us to run the server with the required libraries and dependencies.

4. Storing image on IPFS:
   Initially, we utilized infura to connect to the IPFS and store images for our project campaigns. However, infura later changed their free plans to paid ones, which caused issues with storing images on IPFS. To address this problem, we searched the internet and discovered web3.storage, a service that uses both IPFS and Filecoin to provide an SDK and API for interacting with the system.