Stride in Convolutional Neural Network (CNN)
HTML-код
- Опубликовано: 17 сен 2021
- In this video, we will understand what is Stride in Convolutional Neural Network.
While performing Convolution operation on an image, we move our filter by 1 pixel. It is called having the stride = 1.
But it is also possible to move our filter by 2 pixels or 3 pixels or even more. This is called Strided Convolution in CNN.
Stride in Convolutional Neural Network defines the number of pixels we want to skip or move while performing Convolution Operation in CNN.
➖➖➖➖➖➖➖➖➖➖➖➖➖➖➖
Follow my entire playlist on Convolutional Neural Network (CNN) :
📕 CNN Playlist: • What is CNN in deep le...
At the end of some videos, you will also find quizzes 📑 that can help you to understand the concept and retain your learning.
➖➖➖➖➖➖➖➖➖➖➖➖➖➖➖
✔ Complete Neural Network Playlist: • How Neural Networks wo...
✔ Complete Logistic Regression Playlist: • Logistic Regression Ma...
✔ Complete Linear Regression Playlist: • What is Linear Regress...
➖➖➖➖➖➖➖➖➖➖➖➖➖➖➖
If you want to ride on the Lane of Machine Learning, then Subscribe ▶ to my channel here: ruclips.net/channel/UCJFA
You're underrated! Thank you for making the concepts very clear.
Thankyou so much its helped alot i was confused about stride and paddinh but this playlist is a real problem solver
thankyou bro for the nice explanation. wish you more success
I started watching you from videos of ANN and I benefited a lot. Thank you for your content :)
Your welcome! Glad I could help
GREAT! THANKSSS
too good bro..niceee
thank you u helped me
You forgot the mention why we use STRIDE
good
Do all images have to be converted to a square first?
@coding lane - Can u pls share the PDFs for all the videos in this playlist. Even you u charge fee it is fine. Please share it.
Hi bro! why do we need stride in CNN and how to decide how much to keep for what image? Should stride be neccesarily same in all dimensions? and what effect will it have if we keep it different in different dimensions
Stride reduces the output image size. There is no rule of thumb of keeping a fixed stride... it depends on your network... if your network... if you have big network you might not want to keep higher stride as the output image will be become too small... you can experiment by takinf different stride and see if the model performs better or not... higher stride also means faster computation... which can be helpful.
And regarding choosing different stride for height and width... I think it might not be a good idea... as you will loose more info on the dimensions you are keeping higher stride compared to the other dimension.. so your model might not perform well
@@CodingLane thanks bro!!
You are very talented teacher. I am saying this as your colleague.
Thank you, highly appreciate your words! 🙂
Kindly start making videos on computer vision especially on YOLO.
Sure Waleed... gald you found my videos helpful. I will be uploading more videos on CNN and then explain YOLO as well.
And sorry... won’t be able to upload upcoming video for this week and the next, as I am having a lot of work to deal with. Will add more videos from a week later.
@@CodingLane what I would prefer you is to record 3-4 videos when you have time and then upload them as soon as possible because most students follow your videos on daily basis and if you took a long break that is not helpful for students.
Btw I really like your videos
@@waleedrafi7977 Thank you for your suggestion. Its good to know that these videos help you all. New videos will be out from this week 😇