Aman Dubey

After my 3rd semester, I decided to try my hands on some real-life projects and so, I opted to learn BeagleBone Black for that. I have done this project during my winter traning program at Centre of Excellence at Netaji Subhas Institute of Technology, New Delhi under Prof. D V Gadre.

    This is an IoT based project which uses BeagleBone Black and WhatsApp for its implmentation. Communication between BeagleBone and WhatsApp have been done through Yowsup API. Yowsup is a python library that allows us to use WhatsApp service in Linux systems. It provides most features of an official WhatsApp client in a shell environment such as sending and receiving text messages as well as multimedia messages.

This project focuses on two aspects of smart doorbell i.e. occupant’s end and visitor’s end. Smart doorbell is capable of clicking photos, recording audio and creating an alarm system with notification alerts on WhatsApp having picture and audio of the visitor. This was implemented to allow real time monitoring of the door.
Here BeagleBone Black is connected with Camera, speaker and microphone through USB hub. According to functionality of components work was assigned. When visitor presses the doorbell, camera clicks photo and at the same time user was asked to record 5sec audio. Beaglebone sends both image and audio to occupants WhatsApp. If occupant wants to give some message, he can do so by recording on WhatsApp and user can hear it with the help of speakers.

def credential():
return "91955XXXXXXX","password"
def sendaudio():
    try:
     stack = SendMediaStack(credential(),[(["919545XXXXXX-user-id", "audio.wav"])])
     stack.start()
    except ImageSent as e:
     if ("ERROR" in e.value):
       print("Could not send the image!")

Try this by using RaspberryPi and Beaglebone Black & this will work surely.

This project was developed for Make in India Hackathon organized at IIT Bombay. This is another IoT based project, here I decided to create a prototype for real-time data transfer using Raspberry pi/BeagleBone Black, Android Studio and Image Processing.

    Basic idea behind this project was to provide traffic data statistics alongwith photos of traffic to the user at real-time. Second objective of the project was to control timers using image processing.

Beaglebone will use various image processing algorithms to monitor and control the traffic signal timers accordingly. Eg- More the traffic, more time the green signal will be on for. This will ensure smart flow of vehicles on road.

For Image processing Haar-cascade and Background-subtraction was used:











The second part of the project will focus on a Traffic Forecasting App. Beaglebone will be connected to a server which will keep a track of traffic density with real-time photographs. Server is used to store the data for future reference and also used to provide data to the user on-demand. The user will input source and destination in the app and he’ll be provided with the shortest time taking route along with real time photographs. GPS system will be used to track the user and statistics of the coming traffic will be sent to the user, as per position of user, so that he can change the route accordingly.

Some screen-shots from App are:
    
    

Given below is the flowchart for protocol:

Infosys organized 12-hours long Techzooka Hackathon. Out of many topics given by Infosys for the Hackathon, Smart Parking System was one of the topic which includes Data Analytics, Android App Development, Microcontroller, Image Processing as well as Networking concepts. Infosys spends a lot in parking, so it is important to effeciently manage the space.

    This is also an IoT-based project (like my other projects) & it requires Big-data & Hadoop concepts, Android App, Microcontroller, such as BeagleBone Black & image processing for its implementation

Main objective behind the project was to provide security and deal with capacity management of MULTI-LEVEL PARKING facility available @ Infosys Pune. A prototype was designed having following key-features:

• Employees can register the vehicle details like Registration Number, model, type etc through App & all details are stored in one common server.


• Capturing Number-plate images of each vehicle entering the campus & to check it in the database.


• Raising an alarm for an unidentified vehicle.


• Displaying remaining capacity of parking spaces for each of the vehicle types in App as well as in Hardware.


• Alerting authorities when less than 10% of capacity is available for parking.


• Having specially reserved parking spots for women, handicapped employees and pregnant women.

To implement all these fuctionalities a prototype was developed, which requires personalized login. Now let's understand the concept one-by-one. To capture number-plate, Image Processing Algorithm i.e., OCR(Optical Character Recognition) was used. This Algorithm gives the number in text format, finally this text can be checked in the database of Infosys. For parking system capacity, a central database containing available parking space was managed. App with personalized login will guide the employee to most suitable available place. The database would be updated if employee parks the vehicle. By this way, capacity of parking area was also managed and also authorities can be alerted if less than 10% parking space is available. Personalized login will help people with reserved parking spots.

Some screen-shots from App are:
    
    

Given below is the flowchart for protocol:

Supervised Learning Approach for Precision Agriculture' or in short 'Krishi Kalyan' is a project based on the concept of Precion Agriculture. This project was specially developed for 'Smart India Hackathon' under Ministry of Agriculture and was also appreciated by many Industrialists & Scientists.

    This is also an IoT-based project (like my other projects) & it requires Big-data & Hadoop concepts, Android App, Microcontroller, such as BeagleBone Black & image processing for its implementation. In this project past 100 years of data was trained and forcasted using time series algorithm.

Main objective behind the project was to reduce farmer's efforts and tell them real-time condition as well as suggestions for their field. A prototype was designed having following key-features:

• Analysis & prediction of past 100 years of weather data using Hadoop concepts & storing the results on server.


• Using precision agriculture techniques, real-time soil monitoring system was designed to provide soil conditions.


• By Comparing weather and soil data, most suitable crops and its real-time conditions will be given to the farmer.


• In addition, pest detection model is incorporated in the App.


• Nearby best seller of seeds & pesticide with good reviews and having low price is also shown the farmer.


• A farmer community that will help the farmers to tackle with difficulties that they are facing.

With the soil & weather data, a model is created which will take real time input from the system deployed at the field containing Microcontroller & sensors. Real time soil conditions are given to the model, which predict crops that are suitable for given environmental and soil conditions. Personalized monitoring of the field can be done by using personalized accounts. The App will give all necessary details about the field as well as ways to make farming more efficient. As a future scope, some features like pest detection & nearby best seller can also be implemented by training the data.

Some screen-shots from App are:
    
    
    

Given below is the flowchart for protocol: