So, we have taken the problem statement in the field of Healthcare. The target audience for ReCo.ai is basically the people with cognitive disorders and the elderly. The people having musculoskeletal issues and need physiotherapy. ReCo.ai uses basic IoT tools to perform cognitive tracking and keep the user and the therapist updated regarding the current physical activity the user is performing. This eliminates the need for manual supervision on the patients 24/7 and provides a sense of independence alongside the complete monitored milieu for the guardians. By this AI tool, we aim at scaling it to a much better and more palpable level and introduce it in more future prospects.
Ambient Assisted Living
Ambient Assisted Living (AAL) environments encompass technical systems and the Internet of Things (IoT) tools to support seniors in their daily routines. They aim to enable seniors to live independently and safely for as long as possible when faced with declining physical or cognitive capacities.
The demarcation between traditional and smart systems is automation. A vast number of existing commercial projects are based on predefined rules and actions, which can be changed manually, thereby reducing efficiency. By integrating IoT and Artificial Intelligence, we get a futuristic amalgamation: Ambient Intelligence which has immense prospects to help the ailing.
Visual human activity recognition
Smart environments which include cameras and sensors can be used to obtain images and hence can be used for surveillance and monitoring. This is a viable addition to the project which makes it even serviceable and gives it a commercial use case.
The numerous challenges cannot be enumerated in a small purview but the most prominent ones faced by us during concocting ReCo.ai started with the dataset itself. From the extremely vast library of datasets online selecting the dataset that had enough data points and using just the right features required for curating the model was an extremely tedious task. After that was conquered the next problem started with data preprocessing. The data preprocessing of the data into the right normalization and the right limits eliminating the unusable data involved a righteous amount of patience and expertise. Then on the basis of data points and the limitations of the data involved a lot of mathematical intuition that ended up selecting the right model for the problem statement. After the model was selected the parameterization and architecture that fit the model perfectly over the right number of epochs and the right loss function for evaluation was decided which again involved a lot of statistical intuition. After the final code was curated the backend for the web app was created. Where the guidance of the esteemed mentors was taken into account. The backend code was made by using flask as the functionalities where a lot of errors were encountered. The API was generated and used on flask multiclass classification Cuda and version error were all up to the pinnacle when the frontend came into play as well. Making the attractive front end and the illustrations, logos, and graphics were a recondite subject matter, applying a soothing and minimalist approach by our front end developer used up a lot of creative resources. Developing the frontend with minimal design was also a challenge but we did that too successfully with an eye-soothing design. But after all these challenges were conquered we ended with an AI tool of the future. Presenting… ReCo.AI
Technologies used
Discussion