Authors
ABDELAZIZ MOHAMED, AHMED MOHAMED, KHALED LOAY, ESLAM MAGDY, DR. MUSTAFA ABDULSALAM, ENG. MENNA GAMIL
Publishing Date
28-OCT-2020
Abstract
Nowadays swimming pools is considered one of the main luxuries in the world right now and can be found in clubs, residents, hotels, and schools. Most of these places have a lifeguard to keep an eye on swimmers and rescue them if an accident happens but lifeguards always face difficulties like swimmers who drown in blind spots, So,we propose a system based on IOT technology that classifies swimmers behavior underwater and switches on alerts if the system detects a drowning behavior. The system consists of main 3 nodes. The camera captures videos underwater with microcontroller “nividea jetson”. As a result,Drowning is detected using machine learning, image processing techniques and CNN algorithm. Finally, a mobile message notification will be sent to the medical support account on the mobile application and additionally, a light alarm will alert the pool.
1.1 Background
DROWNING IS THE 3RD UNINTENTIONAL INJURY DEATH WORLDWIDE WITH AN ESTIMATION OF 320,000 DROWNING DEATHS YEARLY MAKING IT A MAIN PUBLIC HEALTH PROBLEM WORLDWIDE, ONE OF THE MAJOR FACTORS THAT CAUSE DROWNING IS AGE AS THE HIGHEST DROWNING RATES ARE AMONG CHILDREN AGED(1-4 YEARS) FOLLOWED BY CHILDREN AGED(5-9 YEARS), DROWNING IS ALSO ONE OF THE TOP 5 CAUSES OF DEATH FOR PEOPLE AGED(1-14 YEARS) IN 48 COUNTRIES. THE RELATIONSHIP OF CHILDREN DROWNING IS ALWAYS LINKED WITH THE ABSENCE OF SUPERVISION ANOTHER MAJOR FACTOR IS THE INCREASED ACCESS TO WATER SUCH AS SWIMMING POOLS [1], NOWADAYS THERE IS 10.4 MILLION RESIDENTIAL AND 309,000 PUBLIC AND PRIVATE SWIMMING POOLS IN THE UNITED STATES ONLY WHICH MAKE A HUGE RISK OF DROWNING AS 79% OF CHILDREN HAVE NO SWIMMING ABILITY[2].
PEOPLE HAVE SHOWN INTEREST TO CREATE SYSTEMS TO HELP IN DECREASING THE RISK OF DROWNING. THOSE SYSTEMS HAVE TWO MAIN TYPES: WEARABLE SENSORS SYSTEM [3], COMPUTER VISION SYSTEMS[4].WEARABLE SENSORS SYSTEMS THAT MADE OF HEADBANDS AND HAND BANDS ARE USED TO MONITOR SWIMMERS SUCH AS HEART RATE SENSOR TO ALERT ONCE THE HEART STOPS BEATING WHICH IS NOT EFFICIENT FOR RESCUE AS IT ALERT IN A LATE STAGE OF DROWNING[5], ALSO PRESSURE DEPTH SENSOR TO ALERT ONCE THE SENSOR REACHES SPECIFIC DEPTH SET BY A THRESHOLD WHICH ALSO NOT EFFICIENT BECAUSE OF MAKING MANY FALSE POSITIVE IF THE SWIMMER IS JUST TAKING A SMALL FAST DIVE[6].
COMPUTER VISION SYSTEMS ARE USED TO DETECT HUMAN UNDERWATER WITH COUNTING THE TIME OF THEM STAYING UNDERWATER UNTIL IT EXCEEDS A SPECIFIC TIME SET WITH A THRESHOLD TO ALERT FOR POSSIBLE DROWNING[7] THIS APPROACH STILL FACING A PROBLEM OF SETTING A STATIC THRESHOLD FOR ALL OF DIFFERENT AGES OF SWIMMERS WHICH IS NOT RIGHT AND CREATES MANY FALSE ALARMS.
THE PROPOSED SYSTEM ENHANCES OLD COMPUTER VISION SYSTEMS. BY ADDING A TRACKING BEHAVIOR FEATURE THAT WILL ALERT ON DETECTING ANY DROWNING BEHAVIOR. THIS INCREASES THE POSSIBILITY OF RESCUING AT EARLY STAGES RATHER THAN DEPENDING ONLY ON A THRESHOLD SET WITH TIME TO LIMIT SWIMMER UNDERWATER.
1.2 Motivation
OUR SYSTEM IS MOTIVATED BY BOTH SYSTEMS OF ABDEL ILAH N. ALSHBATAT AND SHAMMA ALHAMELI. WHO DEVELOPED A COMPUTER-BASED SYSTEM THAT DETECTS DROWNING INCIDENTS IN SWIMMING POOLS BY USING MACHINE LEARNING TECHNIQUES TO TRACK SWIMMERS THAT STAY A LONG TIME UNDERWATER AND
CHI ZHANG, XIAOGUANG LI. THAT CREATED A COMPUTER VISION SYSTEM TO DETECT AND ALERT ANY HUMAN THAT ENTERS A DANGER ZONE THAT IS COVERED BY A CAMERA CONNECTED TO THE SYSTEM.
1.3 Problem Statement
PROVIDE A REAL-TIME ALERT SYSTEM TO INCREASE THE POSSIBILITY OF DROWNING RESCUE BY HELPING LIFEGUARDS TO NOTICE A POSSIBLE DROWNING BY CLASSIFYING THE SWIMMER DROWNING BEHAVIOR USING CAMERA AND MACHINE LEARNING TECHNIQUES.