An Innovative, IoT based, Safe alternative for legacy locker systems
- 0 Collaborators
These days, storing money, gold and other valuables in the bank lockers has become a worrying aspect to the citizens all around the world. According to the statistics on the bank robberies and loots, India almost lost $27.9 million(180 crore rupees) only on loots and burglaries in past 3 years .And there are cases being noticed where the burglars attempted to loot the bank with a disguised costume of a nun on them so as to make the bank managers believe that they are the original owners of the locker. Apart from the incidents happened all around the world, improvisations done to do the lockers safeties every year was mostly found are of only in the mechanical way i.e., lockers were given strength by manipulating the materials used. But, unlike to all those works, we come up with a system that could effectively face these kinds of problems and could even log the data like time of access to locker, changes occurred in the weight of the locker etc., and increase the security of the bank lockers making the individuals feel much safer on their property.As there is no intervention of men it will be a more accurate and safer method.The proposed system works with two levels of security, one of them is face recognition of the owner with the priorly given photo of owner during his registration in the bank.They should pass the face recognition test after which they will enter the second level of authentication where the user has to set the handles to a unique angle key(which is similar that of an ATM Pin) which is provided to them. Our system recognizes the face of the person who visits the bank for access to the locker, by using haar classifier, edge, and line detections features and the faces available in the database and activates the access to the only respective locker. One can noticeably understand that when it is said the person is given access to the locker that means all the other lockers stay deactivated for the access and any trail to open them, triggers the alarm. When the person reaches locker there is a second stage of security, where the person has to open the locker by rotating the handles to a certain angle.This action needs care and can be done perfectly by the owner alone. So, this system could effectively enhance the security of the lockers. ...learn more
Project status: Published/In Market
Overview / Usage
The statistics of bank robberies and burglaries have reached an adverse stage where loot recovered from the incidents was only 20% of the loss. All the other money was exposed to loot and out of those it could be found that the male robbers were almost 5573 in number and female are of 429 in count ,but we can infer from the statistics that due to the complete disguise of the person it is impossible to recognize whether the burglar was a man or woman. So, our system uses a deep learning based face recognition and image detection system that could easily detect many other features unlike the systems that are in use these days. For an instance, if the burglar is carrying a gun like weapon hidden in his overcoat, in normal times there will be minimum observation on these type of things and no individual in the banks could spare time to observe all these things although security systems have been appointed, it makes mistakes which are being rolled in the past events. So, as the first level of the authentication to locker our system recognizes the person and as the algorithm to it is efficient enough to detect the face even in feeble lighting and changes occurred in the face of the person due to the lapse of time can be detected with ease.
Methodology / Approach
Moving to the architecture part of the system designed, the very first thing the user must do is, to enter the unique ID allotted to the person. This ID has significance in the recognition process of the person. This system is designed as such to make use of the S3 bucket in the AWS cloud which is basically a storage service by the Amazon. We programmed a python script in such a way that as soon as the user enters the ID, the camera attached to the locker takes a photo of the user and compare it with the previously stored image of the user which is given during the bank registration process. The uploading of the image into S3 bucket is done using boto3 client which is an Amazon web services accessing package binding for the python and using this boto3 script the picture of the person is uploaded to the aws server and does the face comparison process returning the similarity percentage, confidence of the picture that it is exactly the same as the source picture available in the database. Using a python pull script the data is acquired and processed using data analytics in order to authenticate the user. If the similarity percentage is above 90% the user is allowed for the second level of authentication. The reason why 90% is fixed as the threshold value is that, during the time of capturing of the picture the light ambiance and even if the texture of the skin is dark, the system will be able to extract the maximum features from the face uploaded. So, as to overcome the errors and increase the security the threshold was set to an apt value. In the second level of authentication he has to set both handlebars to the secret angle key. The main advantage of our system is, there will not be much to change with the presently existing locker system as these lockers work on the key basis after which the handles has to be rotated. To adopt our system in present-day locker system set up of two accelerometers are enough. Our system also solves many issues related to customers losing their keys(In this case their face is the main key) , It also solves the issue of not knowing about the status of the locker etc.
Technologies Used
AWS, IoT, Sensors, Data Analytics, Microcontroller, Android.
Repository
https://youtu.be/uOfXwTxBprY?list=PL3uLubnzL2Tml5Nn3IpDjAc097eTpQevy