Sunteți pe pagina 1din 18

INTELLIGENT ALBUM

CREATOR
PROBLEM STATEMENT
 Tocreate an user interface that can take picture(s) from
the user as input and recognize the picture(s).
 The
image recognition can be done using convolution
neural networks.
OBJECTIVE OF THE RESEARCH
 Recognize and classify a photo that the user
uploads.
 Sort the images into respective folders.
 Display the folder name in which the picture has
been placed to the user.
WHY WE USE DEEP LEARNING
They tend to perform well with amount of data wheras old
machine learning models stops improving after a
saturation point.
Feature extraction is done by human in machine learning
whereas deep learning model figure out by itself.
Why CNN

CNN : Pros :
 CNNs can be thought of
automatic feature extractors from
 Feature Extractor
the image.

 No spatial interaction
 While if I use a algorithm with pixel
vector I lose a lot of spatial
interaction between pixels, a CNN  Max Pooling
effectively uses adjacent pixel
information to effectively
downsample the image first by
convolution and then uses a
prediction layer at the end.
Layers

 Convolution Layer
 Pooling Layer
 Flattening Layer
 Connection Layer(Hidden
Layer,Output Layer)
Activation Methods:

 Relu  Rounds off the negative values to


0
 The range will 0 to 1

 Softmax function calculates the


 Softmax
probabilities distribution of the
event over ‘n’ different events.
 The range will 0 to 1
GUI Using Tkinter()

Why Tkinter()

 Python with tkinter outputs the


fastest and easiest way to create
the GUI applications.

 Creating a GUI using tkinter is an


easy task.

 It is a standard Python interface to


the Tk GUI toolkit shipped with
Python.
Output
Problems we faced

 1. Lot of research was done on effective data set for this model

 2. We had a problem to choose the effective batch size

 3. We did trial and error with the number of epoch to get the
necessary accuracy level

 4. To achieve necessary accuracy level we change the size of the


data set and also fluctuated the number of epoch
scope

 The most common application is personal photo organization.

To what extent we can use it?


Future Additions
Objects as dataset
Future Additions

We would add these additional features in future


 • Customize album for junk photos

 • Emotion capturing of the particular picture


Popping up memories at respective festivals

 • Flexible storage of photos

 • Flexible input size


Any queries ?

S-ar putea să vă placă și