The following video represents the last blog and final submission, while demonstrating part of what the developed system can do. It also includes some brief explanations for a more thorough understanding:
To conclude, this project presented an overview of the current smartwatch market and state of the art features already available and researched throughout different fields, especially for medicine and disease prevention. It proceeds to present many different features for health and activity tracking, as well as a watch design, all based directly on the surveys examined.
It provides a complex look into what goes behind the scene in the design of a device that survives on the accuracy of its sensors and the features available to the user and ease of use. With some features going beyond what is expected, in the form of the fully offline voice recognition Convolutional Neural Network, as well as the fusion of 5 different specialised Machine Learning models for weather classification and rainfall regression.
For the future work the following objectives could be completed to fully conlude the project:
Add 6 more voice commands for more granularity in controlling the device outputs. As an example: "What is the temperature?", "What is the humidity?" etc.
Adding "quirky" voice lines to improve the humane feel of the device. For example, if the ML Model classifies the weather as rain, then have a chance to say "Don't forget your umbrella".
Change the voice lines to an alternative sounding even more natural.
Miniaturize the modules and components.
Create the shell 3d model.
Create and test fit the shell by 3D Printing.
Create the shell with a CNC and perform post-processing for a high quality look.
3D Print the support and install electronics.