Pooling Layer in CNN | MaxPooling in Convolutional Neural Network

The pooling operation involves sliding a two-dimensional filter over each channel of the feature map and summarising the features lying within the region covered by the filter.
Code - colab.research.google.com/dri...
Demo - deeplizard.com/resource/pavq7...
============================
Do you want to learn from me?
Check my affordable mentorship program at : learnwith.campusx.in
============================
📱 Grow with us:
CampusX' LinkedIn: / campusx-official
CampusX on Instagram for daily tips: / campusx.official
My LinkedIn: / nitish-singh-03412789
Discord: / discord
👍If you find this video helpful, consider giving it a thumbs up and subscribing for more educational videos on data science!
💭Share your thoughts, experiences, or questions in the comments below. I love hearing from you!
⌚Time Stamps⌚
00:00 - Intro
00:36 - The Problem with Convolution
05:10 - What is translation variance?
07:46 - What is Pooling?
13:05 - Pooling Demo
15:01 - Pooling on Volumes
16:37 - KERAS Demo
18:14 - Advantages of Pooling
22:47 - Types of Pooling
25:36 - Disadvantages of Pooling
27:42 - Outro

[@Abhishek-pb8kt]

Can keep listening to the content for hours without stress or anxiety :D and no drop in enthusiasm or learning. Covering every important detail methodologically.

[@swiss8943]

I wish I could listen to it for hours, but I have my exams right after 2 hours

[@balrajprajesh6473]

First I like, then I watch!

[@artcraftsdecoration8454]

Same here 😀

[@meerzamanatali8914]

Same 😊

[@rishabhvyas856]

One of the best teacher i ever found ❤
I feel so lucky that i found you.
Thank you so much sir i really appreciate you🙏

[@wanderlust_womaniya]

Thank you so much for clarifying this in such a simple language and specially with showing websites that give visual demonstration of such concepts! 🙏🙏🙏🙏

[@yashgupta7604]

Sir you are one of the best teachers i have ever studied from .Keep shinning sir. I am very lucky that i found you.Keep showering knowledge sir 😇

[@AryanSingh-fe5uy]

Sir your teaching flow and method is awesome

[@pravinshende.DataScientist]

I have learned a lot much from this channel .. Thank Nitish sir!

[@chetanmundhe7899]

Am 3rd, I just started watching but thought of putting comment first.... You are greattt broo... Lots of blessings

[@sahuchiragshyamlal3684]

problems with convolution layer:
i) memory issue
ii) translation variance => features are location dependent
solution:
i) strides solve first prob only
ii) pooling solve both prob
relu apply on feature map so non linear feat map then apply pooling(downsampe feature map)
model1.add(MaxPooling2D(pool_size=(2,2),strides=2, padding='valid'))
maxpooling in receptive field(pool_size) keep dominance feature and discard low level details
15:00 min for dimensions
No training require during backpropagation
Advantages: i) reduce size of feat map ii) Translation invariance iii) No need of training bcz its just aggregate opetaion
iv) enhance feat only in max pooling
24:00 global max and avg pooling
Disadvantages:
i) Translation invariance not used in Image segmentation
ii) Loose lot of info

[@debojitmandal8670]

But from wht i understood is that pooling will reduce your image size therefore your loosing some information how do you solve that

[@piyushpathak7311]

Sir please complete this series as soon as possible..

[@AlAmin-xy5ff]

You are really awesome sir!!!!!!

[@savyasachi6988]

very nice pooling explanation

[@AmitKumar-kt8oc]

Amazing 😍

[@algocoholic2494]

2nd time.... 1st view and 1st comment 🤣🤣🤣🤣🤣...love you sir.... Hope you have read the form that I filled

[@homeboy1733]

Awesome Sir

[@mr.deep.]

Best

[@pravinshende.DataScientist]

every time you properly understand the why concepts that why we get the concptual clarity .. so ratta marne ki jarurat nahi hoti

[@narendraparmar1631]

help a lot
Thanks

[@guru_bro]

Very good 👍

[@Pawan-pm4tw]

Salute sir❤️

[@gurpreetkaur-pf1bf]

Thku from heart ❤️

[@tanveerbashir8393]

Thank you🥰😍😇

[@vivekacharya3652]

Main first😄😄

[@sandipansarkar9211]

finished watching

[@ppal6329]

The best

[@elonmusk4267]

great

[@mihirthakkare504]

Santre colour ka billa❤ in video is lub

[@rb4754]

First I like then I watch the video and then I comment...

[@yashjain6372]

awesome video.
Doubt : 20 Min
But why we are not getting shape of 8 by pooling ?

[@nipunsingh7740]

If possible, do upload your notes too along with the videos

[@farhadkhan3893]

you discussed a problem in the previous video of Padding and strides that some information is lost while featuring the images so you applied the padding to maintain the size of the image. so my question is, isn't it a problem that some information is lost in pooling?

[@MARTIN-101]

in pooling information is merged. in padding extra information is added.

[@muhammadyaseen2856]

Sir please create playlist on -- Computer Vision and Image Processing --

[@rahuljha3686]

✓ done

[@divyakarlapudi]

best

[@vairab1876]

What is the size of kernel in conv2d_1 ? there are 32 kernel , is it (3,3) or (3,3,32) ?

[@rachitsingh4913]

Can we use padding and pooling together ? Because by using padding we are actually stopping our features from loosing information whereas by using pooling we are loosing some information . Can someone please clear my doubt ?

[@divakarsaraswat7820]

Sir, Data analyst k liye full road map machine learning ka

[@chetanmundhe7899]

Quick question... Do you work for any company or work for yourself?

[@ebrahimhaquebhatti3258]

2:50 Isn't it 224x224x3?

[@harshumokal5736]

Please do something on object detection

[@MN-gn7lt]

Sir course jaldi launch karo na