As can be seen in the diagram, the system has three parts: In-car devices, On-cloud components and User device.
In more detail, the in-car devices include microphone, Raspberry Pi, Temperature Sensor, Intel Edison and Fan. On-cloud components include Amazon ML and EC2. The User device is an app running in Android.
The whole running process is as follows:
(1) Firstly, microphone located in Raspberry Pi and temperature sensor located in Intel Edison are continuously recording sound and temperature inside the car. Meanwhile, the Raspberry Pi can send the sound recorded to machine learning part running in Amazon ML and get back the ML results ("1" represents danger and "0" represents safe). And they will send their results to the server located in AWS EC2.
(2) The machine learning part is trained by our recorded sound set that is introduced under " DETAILED DESCRIPTION" -"REAL DATA COLLECTION" part. If the machine learning judges a baby crying in the car. When dangerous situation is detected, the APP can receive alarm messages sent by EC2 server, notifying the danger and current temperature in the car.
(3) After receiving the message, the user can open the fan through APP as a counter measure. Furthermore, a SNS notification will be sent to the phone. The SNS notification and APP alert forms a dual fail-safe.
(4) The fan will run different INTENSITY according to different temperatures inside the car. Here, if temperature> 20 centigrade, it will run stronger. Furthermore, the fan will run 60 seconds each time.
In more detail, the in-car devices include microphone, Raspberry Pi, Temperature Sensor, Intel Edison and Fan. On-cloud components include Amazon ML and EC2. The User device is an app running in Android.
The whole running process is as follows:
(1) Firstly, microphone located in Raspberry Pi and temperature sensor located in Intel Edison are continuously recording sound and temperature inside the car. Meanwhile, the Raspberry Pi can send the sound recorded to machine learning part running in Amazon ML and get back the ML results ("1" represents danger and "0" represents safe). And they will send their results to the server located in AWS EC2.
(2) The machine learning part is trained by our recorded sound set that is introduced under " DETAILED DESCRIPTION" -"REAL DATA COLLECTION" part. If the machine learning judges a baby crying in the car. When dangerous situation is detected, the APP can receive alarm messages sent by EC2 server, notifying the danger and current temperature in the car.
(3) After receiving the message, the user can open the fan through APP as a counter measure. Furthermore, a SNS notification will be sent to the phone. The SNS notification and APP alert forms a dual fail-safe.
(4) The fan will run different INTENSITY according to different temperatures inside the car. Here, if temperature> 20 centigrade, it will run stronger. Furthermore, the fan will run 60 seconds each time.