Hang tight while we fetch the video data and transcripts. This only takes a moment.
Connecting to YouTube player…
Fetching transcript data…
We’ll display the transcript, summary, and all view options as soon as everything loads.
Next steps
Loading transcript tools…
Deep learning project end to end | Potato Disease Classification Using CNN - 1 : Problem Statement | codebasics | YouTubeToText
YouTube Transcript: Deep learning project end to end | Potato Disease Classification Using CNN - 1 : Problem Statement
Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.
Share:
Video Transcript
Video Summary
Summary
Core Theme
This content introduces an end-to-end deep learning project series focused on building a mobile application to detect potato plant diseases, covering the entire development lifecycle from data collection to deployment.
In this video we are beginning an
end-to-end deep learning project series
in agriculture domain.
The series will have total seven to
eight videos in which we'll start with
data collection first and then we'll
look into model building. We'll also look
into some of the ML Ops using TF serving.
We will build our backend server using fast
API and then we'll deploy the model to
Google Cloud or GCP and we'll have
Google Cloud functions running on top of
it and that those functions will be
called by
a mobile app
written in React Native so it will be an
end-to-end application very useful in
agriculture
domain. So let's first look at the
problem statement.
[Music]
Farmers who grow potatoes are facing lot
of economic losses every year because of
various diseases that can happen to a
potato plant.
There are two common diseases
known as early blight and late blight
early blight is caused by a fungus and
late blight is caused by a specific
microorganism
and if a farmer can detect these
diseases early and apply appropriate
treatment then it can save lot of waste
and prevent the economic loss.
The treatments for early blight and late
blight are little different so it's
important that you accurately identify
what kind of disease is there in that
potato plant.
AtliQ Agriculture is an AI company that
happens to solve problems in agriculture
domain and that company has taken this
project and they have decided to build a
mobile application which they can give
it to a farmer. And farmer all they need
to do is go to their farm and just take
a picture of the plant and the mobile
application will tell them whether the
potato plant is healthy or it has one of
these diseases and behind the scene it
will be using deep learning and
convolutional neural network. Now you are
a data scientist working for AtliQ
Agriculture and your job is to work on
this project and to build this whole
application end to end.
So first you will gather in a room along
with your colleagues and you will start
discussing the technical architecture.
I will take a quick break and talk about
AtliQ technologies which is my
brother software development company do
you need help with mobile app
development or data analytics solution
or you want to do digital marketing for
your business you can take AtliQ
technology services they have so many
international clients and I guarantee
you won't be disappointed with that
service. Just click on the contact button
here fill out the contact form or call
them directly. Let's get back to our
video now.
Any supervised machine learning project
starts with data collection. Data that
you can use as a training data set. In
our case we need to collect images of a
healthy potato plant leaf and the potato
which has early blight or a late blight
disease.
So we'll cover how exactly we're going
to do that but let's say you have
covered all these images
then comes data cleaning and
pre-processing step for which we will be
using tf data set and data augmentation.
Data augmentation because we might not
have enough diverse set of images so we
need to rotate and flip and you know
adjust contrast to create
more
training samples.
Once we have that we'll use
model building using convolutional
neural network. CNN is a standard way of
uh doing you know image classification
as of 2021 and therefore we'll be using
CNN and then export the train model
onto our disk.
Then we'll cover
some of the ML OPs concept using TF
serving where we'll have a TF server
serving server running on top of these
exported models which can solve you know
different versions of
these models and tf serving will be
called from fast API.
Now you will be asking why we need fast
API and DF serving both well. Don't
worry we will be going over all those
things we can have just a fast API
server without using TFF serving. We'll
cover both the approaches and you will
understand the benefit of having TF
serving in addition to fast API server
so I will be filling the layers of onion
in a way that
you don't get to cry basically. So trust
me on that
then we'll build a website in React JS
React.js is a hot technology as of today
for doing a
website app development and that will be
calling
the fast API server where you can drag
and drop the image and it will tell you
the label. Whether it's a healthy early
blight or late blight
now comes the interesting phase which is
the mobile app development. So first
we'll do website development. We'll test
things out locally on our computer and
when we are ready to go then we move
on to mobile app development. For that
we keep the architecture same up till
this point but then we
convert these exported
float models into TF lite model using
quantization. Now I did a video on
quantization, TF light all of that so you
can watch that video, but quantization is
a way to
reduce the size of your model so that
your model is occupying less memory. You
can deploy it on cell phone on edge
devices and also
the inference speed is much faster
once we have exported TF
lite models then we will deploy those to
and will write Google Cloud functions
which are similar to AWS lambda if you
have heard about serverless architecture
and AWS Lambda so google cloud functions
are kind of
like that and then these functions will
be serving our cool mobile application.
So see here in this mobile application
you take a picture of your potential
plan leave and our application will do
the prediction it will tell you if it's
a healthy or diseased plant and this
application will be written in react
native which is a hybrid mobile app
development framework. Now you might be
having a question why didn't we deploy
TF lite model directly to mobile app?
well,
we will cover that maybe in the future
video but for this time I want to deploy
this model to Google Cloud
and just kind of show you how this whole
things works. So yeah,
the ideal way is to deploy tflight model
to your mobile app directly. I tried it
but I was facing some issues so I went
for a different architecture but in
future when those issues are resolved I
might
show upload a different video you know
where I'll deploy TF flight model
directly to mobile app. So overall in
terms of technology stack for model
building we use tensorflow CNN data
augmentation tf data set. And by the way
I have videos on all these topics so if
you don't know any of these you can
watch it as we go in this
project.
At every step I will tell you the
prerequisite like what videos you need
to watch and so on so don't worry even
if you don't know about all these
jargons you will be able to grasp this
with this project really well. I promise
that.
As a back-end server and ml ops will be
using TF serving and fast API
in terms of model optimization we are
using quantization along with tensorflow
lite and for front end and deployment we
are using dxjs for website React Native
for mobile app development and we are
using GCP for our deployment.
As a prerequisite you need to have a of
course basic knowledge of Python if you
don't you can go to Youtube search for
code basic Python tutorial playlist
and in this playlist
at least watch first
14 or 15 videos and that should be
sufficient.
Another thing you need is a deep
learning basic knowledge for that go to
Youtube Codebasics deep learning
playlist.
This is a big playlist you don't need to
watch all the videos. But watch till
I would say video number 24 image
classification using CNN and that way you
know the basics of neural network.
You know CNN and so on in here you can
ignore video number. Let's say 18 19 20
or even
I would say
video number
14 to 20 something. If you don't watch
it's still okay but remaining videos
please watch it so that you have your
basics covered. Once the technical
architecture is decided the next step in
the second video would be to look into
data collection and data pre-processing
techniques.
If you are liking this video series so
far please give it a thumbs up and share
it with your friends these kind of
end-to-end projects look pretty good on
your resume. If you're trying to get a
job as a machine learning engineer or a
data scientist you can do this kind of
projects. Maybe take my project and
customize it maybe instead of potato try
to identify the diseases in the tomato
plants and I will give by the way those
exercises which will be super cool so
make sure you watch the entire series
and share it with your friends who wants
to learn end-to-end deep learning
projects.
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.
