Prof briefing 1

[leon-thomm]

1. narrow down the use case - what state/s exactly do we want to predict, and how?
2. notice the additional accelerometer in the set, it is much lighter than the phone
3. it's ok to pull in additional data sources to feed the models, such as OpenWeatherMap API
4. architecture seems good
5. finish POC in two weeks from now

[leon-thomm]

1. narrow down the use case - what state/s exactly do we want to predict, and how?

One easy solution would be diving and beach management. Data such as visibility, temperature, and water turbulence is of course very useful for diving activities. As an extension of that, the data could be used to decide whether to close the beach on certain days, e.g. water is too cold, water too turbulent (high current), sharks recorded, etc.

[cheehongw]

narrow down the use case - what state/s exactly do we want to predict, and how?

babies in water

In homes with swimming pools, there is a risk where young children may wander off and fall into the pool without adult supervision. A wave detection system, coupled with cameras can be used to detect if an object (in this case, a child) has fallen into the pool.

With a buoy and some learning on the cloud, we can distinguish between:

1. periodic wave movements due to wind / still water versus
2. wave motion generated by a thrashing living thing.

Upon detection, we can actuate an alarm in the house to notify someone who can help. Multiple buoys can be used to span a larger pool to increase coverage of detection area.

[leon-thomm]

We are deciding on diving forecast.

parameters

• visibility
• water turbulance
• temperature

Idea: combine and classify our underwater predictions with above-water predictions from open APIs to generate an aggregate quality score.

final proposal

Our buoys collect underwater measures like visibility, turbulance, and temperature, and upload the data to the cloud. We then combine the data with above-water weather forecast from open APIs in order to provide divers with a quality score for diving activities, so divers can plan better based on the conditions forecast.

Note that the data is obviously useful for other applications, such as beach management, and the project could be extended into these use cases, but for now we focus only on the quality score for diving activities.

[exetr]

[leon-thomm]

green light from prof

Yes this sounds good to me. Your groups idea is a lot more clear now. My question for you is: what is a diving quality score? Is this something that is available or already defined? Regardless, make sure you select a limited number of such classes/scores. Think about how you will collect such data. Data processing diagram looks fine.