Professional Documents
Culture Documents
A Novel Approach Towards Interface For Touchless Scanning of A Biodiversity Thing
A Novel Approach Towards Interface For Touchless Scanning of A Biodiversity Thing
Abstract – Huge number of Biodiversity Combined with other tech trends that will
things are available on the earth. Some of play a large part in what we’re talking about
things are known to us but some are very in 2018, the touch less interface will move
rare or creepy that we don’t know about us far beyond the tap and swipe.
them. So, to get detail about all the
biodiversity things we need to take help
from technology. Artificial Intelligence II. PROBLEM STATEMENT
makes this task very easy. With the help of It is a well known fact that there are many
image recognition and speech recognition species of biodiversity in the world. So it is
we make easy to getting the details about very difficult to know about all the names
biodiversity species or things. and details about these species and specially
Keywords - Biodiversity, Convolution the species which do not belong to your
Neural Networks country or new,we can analyze them and
can gather information about them through
this.So we need to use artificial intelligence
I. INTRODUCTION for know every details about species.
The user interface is key to our experiences
with our tech gadgets. However most of
III. BASIC IDEA-SCAN
these gadgets rely on some kind of physical
ANYTHING AND LET YOUR
interaction in order to work. This will
PHONE DO THE REST
change in 2018, with a focus on the “touch
less” interface. [1] Many people rely on their smart phones to
search for things online. At the movies,
Imagine picking up your empty juice bottle
users might try to identify an actor from a
and saying “Siri, order this.” Amazon
film trailer. At a concert, they might hear a
Alexa, Siri, and other virtual assistants have
song and check which album it was on.
already begun to train us to no longer rely
When shopping, they might try to find the
on our thumbs and fingers. But as we noted
best deal on a product by searching nearby
in our discussion of the future beyond a
stores. Apps that identify songs, images,
Smartphone there are other ways to interact
and video, or that read barcodes, make it
that could be useful.
easier to do this. New discovery: With a
Imagine picking up your empty bottle of new app from Dig marc, someone could
juice and saying “Siri, order this” and it scan photos from a newspaper or other
being ordered for you without you since media to pull up more information—
Siri understands what you’re looking at or provided those images are in the company’s
holding. Other features, like customized database.
responses based on who is asking, are
A single app designed to identify input
already available and will only become
from a person’s environment and pull up
more prevalent throughout the year.
related information.
Similar to like Barcode Scanner Discover online and brings up related information on
uses a smart phone’s camera and the user’s phone. While Discover’s image
microphone to “capture” a sample of audio and video search functions only work with
or an image, then identifies it through the company’s own digital watermarking
Digimarc’s own database and searches for system.
related material online.
(photo search), and perform object The Convolution layer is always the first.
recognition within scenes. They are Тhe image (matrix with pixel values) is
algorithms that can identify faces, entered into it. Imagine that the reading of
individuals, street signs, tumours, the input matrix begins at the top left of
platypuses and many other aspects of visual image. Next the software selects a smaller
data.[5] matrix there, which is called a filter (or
neuron, or core). Then the filter produces
Let us consider the use of CNN for image
convolution, i.e. moves along the input
classification in more detail. The main task
image. The filter’s task is to multiply its
of image classification is acceptance of the
values by the original pixel values. All
input image and the following definition of
these multiplications are summed up. One
its class. This is a skill that people learn
number is obtained in the end. Since the
from their birth and are able to easily
filter has read the image only in the upper
determine that the image in the picture is an
left corner, it moves further and further
elephant. But the computer sees the pictures
right by 1 unit performing a similar
quite differently:
operation. After passing the filter across all
positions, a matrix is obtained, but smaller
then a input matrix.
This operation, from a human perspective,
is analogous to identifying boundaries and
simple colours on the image. But in order to
recognize the properties of a higher level
such as the trunk or large ears the whole
network is needed. The network will consist
of several convolution networks mixed with
nonlinear and pooling layers. When the
Figure - 2 image passes through one convolution
layer, the output of the first layer becomes
the input for the second layer. And this
Instead of the image, the computer sees an happens with every further convolution
array of pixels. For example, if image size layer.
is 300 x 300. In this case, the size of the
The nonlinear layer is added after each
array will be 300x300x3. Where 300 is
convolution operation. It has an activation
width, next 300 is height and 3 is RGB function, which brings nonlinear property.
channel values. The computer is assigned a Without this property a network would not
value from 0 to 255 to each of these be sufficiently intense and will not be able
numbers. Тhis values describes the intensity to model the response variable (as a class
of the pixel at each point.
label).
To solve this problem the computer looks The pooling layer follows the nonlinear
for the characteristics of the base level. In layer. It works with width and height of the
human understanding such characteristics image and performs a down sampling
are for example the trunk or large ears. For operation on them. As a result the image
the computer, these characteristics are volume is reduced. This means that if some
boundaries or curvatures. And then through features (as for example boundaries) have
the groups of convolution layers the
already been identified in the previous
computer constructs more abstract convolution operation, than a detailed
concepts. In more detail: the image is image is no longer needed for further
passed through a series of convolution,
processing, and it is compressed to less
nonlinear, pooling layers and fully detailed pictures.
connected layers, and then generates the
output.
Figure - 3
After completion of series of convolution, We take an example from Biodiversity
nonlinear and pooling layers, it is necessary where there are some flowers, plants, birds
to attach a fully connected layer. This layer we rarely see. We have never seen them and
takes the output information from they do not belong to our country. We want
convolution networks. Attaching a fully to get information about them so in that case
connected layer to the end of the network this technique can be used.
results in an N dimensional vector, where N
We use our device as scanning a
is the amount of classes from which the
Biodiversity thing, suppose we pick up a
model selects the desired class. A fragment
fruit and asking from smart phone while
of the code of this model written in Python
holding the fruit "What is this"?
will be considered further in the practical
part.[6] After scanning the smart phone searching
this thing from the data base and detecting
from data base. After detecting we get detail
V. PROCESS information about that thing.
VI. BENEFITS
Touch less scanning feature is beneficial for
those who wants to learn and know about
things related to biodiversity.
VII. CONCLUSION
IX. REFERENCES
[1]. https://www.aivanet.com/2018/01/5-
tech-trends-youll-be-talking-about-in-
2018
[2]. https://www.technologyreview.com
/s/425907/scan-anything-and-let-your-
phone-do-the-rest/amp/
[3]. https://www.technologyreview.com/
s/415513/touchless-3-d-fingerprinting/
amp/
[4]. https://www.smartsheet.com/voice-
assistants-artificial-intelligence
[5]. https://skymind.ai/wiki/convolutional-
network
[6]. https://medium.com/@ksusorokina/
image-classification-with-
convolutional- neural-networks-
496815db12a8