Anyone else became vastly amused and super excited when figured out its Mandy's video after a while? LOOOOL She has extremely relaxing voice and face. Good for Chris!!
great video, liked this demo so much! it helped me understand the filters and their role much better :) you're an excellent teacher 🥰 and it's nice to see who's behind the camera! (I'm currently "binge" watching the DL fundamentals playlist haha)
Glad to hear that Bildad! Thank you. Btw we've responded to your email but haven't heard back from you. Check your spam just in case you haven't yet seen it.
Congratulations on your 100K subs, Wishing you guys 100M and Beyond :) I have been a Subscriber ever since this channel was 40kish. I have learnt all the basics of DL from this channel before diving deep into complex concepts likes GANS or Transformers. I really thank you guys for explaining all the concepts so well yet with so much ease. I wish more people knew about this channel ( and BTW I always recommend my college mates to learn the basics of DL and CNN from here whenever they ask me for DL resources. LoL)
Thanks! This vedio was much needed! Also because of you I have been able to do my minor project on mask detection , its still in progress but will complete it soon, you make learning so easy, I have decided I will learn any thing you will teach 😌😌
extraordinary work ! and i think if you guys tell something about the filter size relative to image size that would be a critical point to mention because on large images these small filter not only does not work properly but sometimes they decrease the performance.
Good teaching material! One question though. Where did you get those filters from? I actually see those filters in every tutorial but have no clue how they are created?
Do we tell CNN to detect top edge or left edge? Is there a top edge detecting filter? In convolutional neural networks (CNNs), there is no explicit instruction to detect top or left edges. Instead, CNNs learn to detect various types of edges and features automatically through the training process. CNNs typically start with a set of convolutional layers, where each layer applies a set of learnable filters (also called kernels or weights) to the input data. These filters are initially randomly initialized and are then updated during the training process to detect specific patterns or features in the input data. There is no dedicated "top edge detecting filter" or "left edge detecting filter" in CNNs. However, during the training process, some of the filters in the early convolutional layers tend to learn to detect low-level features such as edges and corners, regardless of their orientation. The orientation of the learned edge detectors depends on the input data and the task at hand. For example, if the input data consists of images with a predominant horizontal or vertical structure (e.g., text lines, buildings, etc.), the CNN is likely to learn filters that detect horizontal or vertical edges, respectively. It's important to note that CNNs are designed to be translation-invariant, meaning that the learned features should be able to detect patterns regardless of their position in the input data. This is achieved through the use of convolutional operations, where the same filter is applied across the entire input data, and through the use of pooling layers, which introduce spatial invariance. In summary, while CNNs do not have explicit instructions or dedicated filters for detecting top or left edges, they learn to detect various types of edges and features, including top and left edges, through the training process and the learned filter weights in the convolutional layers.
Thankyou ma'am for your beautiful and excellent explanations.... Maam please can you help me to build a basic cnn model for language translation... Please Maam i am stuck in some part of the code i have ....
You mention the values of the filter are random. From what I see, they are Not random, but each has a pattern. Also, earlier [paid] lessons say fully connected first layer reads a flattened dataset, but a CNN requires the dataset to be in the original rows and columns. [edit: or at least to recognize that structure. A flattened dataset by itself does not preserve local spatial features.]
Hey there Mandy, you are an excellent teacher. I had saw your keras tutorial in freecodecamp youtube channel which was awesome. I have a doubt regarding cat vs dog classifier which you have explained us as an example. When i tried to classify with dataset face vs noface from same algorithm, it gives very bad accuracy of 50%. Would you please explain me is this approach is whether right or not For face classification. How should i approach face classification. please make a video on face classification problem too. And give some insight about this matter on my comment. Thank you. Btw you look as pretty as gal gadot.
👉 Experiment with the interactive convolution demo app yourself!
🔗 deeplizard.com/resource/pavq7noze2
👀 CHECK OUT OUR VLOG:
🔗 ruclips.net/user/deeplizardvlog
I'm so happy that you guys are back now lol
Anyone else became vastly amused and super excited when figured out its Mandy's video after a while? LOOOOL
She has extremely relaxing voice and face. Good for Chris!!
😅
Last time I watched your video I was still in college. Where have been so long? Glad that you guys are back now!
Good to see you back after 7 months 😀
What a beautiful way to demonstrate the concepts, it made everything click for me! Very well done.
My fav vlogger ♥️♥️. I graduated from my masters degree. Your videos helped me through it all.
Ahh congrats Sanni! 🥳🥳
Brilliant idea and excellent work to put up such intuitive demo! Thank you deeplizard!
Wow ! Awesome application 👏 !
By the way the background at 12:40 was so wonderful !
Thank you! We thought the sky during that time was beautiful too!
Thanks for putting so much time and effort to create such beautiful tutorials.
Thank you very much, Madam. Best illustration video I have seen so far on Convolution Filters.
Thank you so much... This is one of the best videos ever seen....
great video, liked this demo so much! it helped me understand the filters and their role much better :)
you're an excellent teacher 🥰 and it's nice to see who's behind the camera! (I'm currently "binge" watching the DL fundamentals playlist haha)
Thank you!
Thank you for this tool!
Awesome! what a method to present/teach the CNNs in a easiest style. I appreciated
I love the course due to her voice!!! It is exremly loveable
Best Explanation 🎉
Amazing apps with great explanation ,
thank you very much 👍👍👍
After 7 months.. Thanks! 💚💚
Very very simple. I mean, it is not simple but you made it a very simple topic.
You teach so well.
Thanks
Glad to hear that Bildad! Thank you. Btw we've responded to your email but haven't heard back from you. Check your spam just in case you haven't yet seen it.
was watching a playlist and thought this was AI generated voice. Good to know its a real human :)
Congratulations on your 100K subs, Wishing you guys 100M and Beyond :) I have been a Subscriber ever since this channel was 40kish. I have learnt all the basics of DL from this channel before diving deep into complex concepts likes GANS or Transformers. I really thank you guys for explaining all the concepts so well yet with so much ease. I wish more people knew about this channel ( and BTW I always recommend my college mates to learn the basics of DL and CNN from here whenever they ask me for DL resources. LoL)
Thanks so much Zafer!
You save my life with this tool! thank you!
Brilliant, Thank u!
You are phenomenal. Thank you. Incredibly helpful.
Thanks! This vedio was much needed! Also because of you I have been able to do my minor project on mask detection , its still in progress but will complete it soon, you make learning so easy, I have decided I will learn any thing you will teach 😌😌
Great job!
Smart and gorgeous lady, thank you so much!!
Thank you Mandy!! This is such wonderful way to show convolutions!
Awesome tool for teaching convolution. Thanks
Nice explanation
Great, thank you
GREAT JOB
see who is back yay
So glad y’all are back!!!
Thank you for carrying my Keras, Pytorch and deeplearning theory skills lol
So happy that you are back! Good luck!
extraordinary work ! and i think if you guys tell something about the filter size relative to image size that would be a critical point to mention because on large images these small filter not only does not work properly but sometimes they decrease the performance.
Andy been lifting 💪🏾
😁💪 www.youtube.com/@deeplizardlifts
You are a beautiful soul. God bless you !!
Nice work! Make it easy to understand!
Nice to see you again.
welcome back
This is so awesome, Thank you very much.
Good teaching material!
One question though. Where did you get those filters from? I actually see those filters in every tutorial but have no clue how they are created?
Thanks for this. Also, soooo close to 100k!!
🥳🥳
I like your videos, and I think it is not too early to present transformers, attention and self-attention on your channel.
This is so helpful
Do we tell CNN to detect top edge or left edge? Is there a top edge detecting filter?
In convolutional neural networks (CNNs), there is no explicit instruction to detect top or left edges. Instead, CNNs learn to detect various types of edges and features automatically through the training process.
CNNs typically start with a set of convolutional layers, where each layer applies a set of learnable filters (also called kernels or weights) to the input data. These filters are initially randomly initialized and are then updated during the training process to detect specific patterns or features in the input data.
There is no dedicated "top edge detecting filter" or "left edge detecting filter" in CNNs. However, during the training process, some of the filters in the early convolutional layers tend to learn to detect low-level features such as edges and corners, regardless of their orientation.
The orientation of the learned edge detectors depends on the input data and the task at hand. For example, if the input data consists of images with a predominant horizontal or vertical structure (e.g., text lines, buildings, etc.), the CNN is likely to learn filters that detect horizontal or vertical edges, respectively.
It's important to note that CNNs are designed to be translation-invariant, meaning that the learned features should be able to detect patterns regardless of their position in the input data. This is achieved through the use of convolutional operations, where the same filter is applied across the entire input data, and through the use of pooling layers, which introduce spatial invariance.
In summary, while CNNs do not have explicit instructions or dedicated filters for detecting top or left edges, they learn to detect various types of edges and features, including top and left edges, through the training process and the learned filter weights in the convolutional layers.
The star wars hummed theme during the sped up calculations got me dead 😂🛸
🌌😆
And they're back. . .
🥳
Excellent demo!
Thank you madam for such an amazing tutorial. Thanks.
Welcome back awesome app!
Love it, thanks deeplizard
Beautiful!
deeplizard is back! Yayy
this is a w e s o m e !!
Thankyou ma'am for your beautiful and excellent explanations.... Maam please can you help me to build a basic cnn model for language translation... Please Maam i am stuck in some part of the code i have ....
luv ur vidoes 🤟😍😍
Deep lizard >>
good explation
You mention the values of the filter are random. From what I see, they are Not random, but each has a pattern. Also, earlier [paid] lessons say fully connected first layer reads a flattened dataset, but a CNN requires the dataset to be in the original rows and columns. [edit: or at least to recognize that structure. A flattened dataset by itself does not preserve local spatial features.]
Would you mind making a video on how to build your own RNNCell implementation for keras? I'm struggling to find a documentation
Nice
Capsule net for next Yeey
Please do a c
Video on implimetation of Mobile net V3..(image classification)
Thanks in advance
How does this work with RGB images?
Mam can u make more videos for nlp
What activation function is being used? How are we allowed to get high negative values? Thanks a lot!
This demo shows the convolution operation with no activation.
Hey there Mandy, you are an excellent teacher. I had saw your keras tutorial in freecodecamp youtube channel which was awesome.
I have a doubt regarding cat vs dog classifier which you have explained us as an example.
When i tried to classify with dataset face vs noface from same algorithm, it gives very bad accuracy of 50%.
Would you please explain me is this approach is whether right or not
For face classification.
How should i approach face classification.
please make a video on face classification problem too.
And give some insight about this matter on my comment. Thank you.
Btw you look as pretty as gal gadot.
lovely accent
I want to be an agent and explore her environment to get maximum reward after each episode and never update my Q-table . 😎
Where did you came from after 6 months?!!!
🙈
i keep staring at the t-shirt
she is so beautiful... confused what to see her or demo
you are so beautiful 😅😅😅
nice but too much talking.