We are students of Amity University Mumbai who wish to bring about a positive change in the lives of the people around us and provide them value by producing software which helps increase their ease of living.

Understanding the Problem

Identifying the arch type of a foot is crucial in preventing foot-related injuries, improving athletic performance, and selecting appropriate footwear. Manual analysis of arch type is not only time-consuming but also prone to errors. Hence, an AI-based solution that automates this process is essential and would offer significant benefits.

Planning Our Approach

To establish a pool of potential solutions to work with, we conducted in-depth study and examined numerous theories, articles, and papers. We put some of the most promising ideas to the test and recognised both their advantages and disadvantages.

alt

Robust, Structured, Interactive and Executable

Smartly Coded & Maintained.

The first step to any successful project is planning and setting guidelines to follow. We also made sure to define our pathways before investing valueable time and resources into our project.

Choosing the right training model.

Choosing the right training model is necessary because it directly impacts the accuracy and effectiveness of the resulting model. A well-chosen model can learn from data and make accurate predictions, while a poorly chosen model can lead to inaccurate or unreliable results.

The choice of model depends on the nature of the problem to be solved, the type and amount of data available, and the specific requirements of the application. Hence, after going through different algorithms, we decided to have a Neural Network for this classification problem.

Current Functionalities

Segmentation of the Foot area– A model to crop the image to fit the focus area. (50% - 60% Accurate)

Classification over a given test set.

Including User defined image in the test set.

What can we achieve

Our Proposals

Research
We conducted extensive research, explored multiple theories, articles and papers to find out a pool of possible solutions to work with. We tested few of the most promising solutions which were feasible by time and understood both the pros and cons of each.
Choosing and Training the model
Of the models we found relevant to the problem statement, we shall train a couple and test their performance on a set of custom inputs. Models which provide the best and relevant responses will rank higher amongst the set.
Adding Functionality
Consolidating our website to encapsulate both the functions of the frontend and the backend, integrating the deep learning algorithm into the system, setting up data flows and user experience.

Team

Gangs Of Wasseypur