I can't explain in my word. Really mam you are best teacher😊 on the youtub. But i don't know why your subscribers is very less on TH-cam.😔😔, I will share you channel with my friends.👍😊
Your videos are very informative but I have suggestion for you that your picture block covers half of ppt, can you reduce your picture block or you can use green screen. Thanks
Hi, Thanks for the amazing explanation. I have a question and would request your answer: If alpha beta and gamma are fixed. φ is obtained using Grid search. Here I want to understand, that during grid search we have to consider different values of φ. Also, I want to understand what is f here. How does the value of f contribute to the Neural Network? Please Advise. Thank You Sumit
at 32:36 you said the authors already fixed the values of depth width resolution factors, and found phi value, but in that research paper they kept the value of phi as 1 and did a grid search for alpha beta gamma after they found the values they adjusted the phi to upscale different Efficientnets, correct if i am wrong it will be helpful? and the video is good
Maam do you have accumulative explanation on all the CNN models so that a person can get idea of which CNN model has research gap and needs to be worked on it compare to other cnn models on PhD level research gap.
Grid search is Hyperparameter tuning technique, researcher would have have done investigation here by trying multiple values and finalized the best or optimized values as depth=1.20;width=1.10 and resolution=1.15
Hello, I am using pre trained EfficientNetB0 with ImageNet, I have costume 166 image set, I am using regularizations(drop out and batch normalization )/Augmentation technique. But Getting pretty bad result my val_loss increases and val_accuracy remain constant. could you please help.
Hello Ma'am, Can we perform Detection of Malware in windows system (output expected is only in YES/NO) using EfficientNet, or is it only meant for image classification?
Mam, alpha=1.2,beta=1.1& gamma=1.15 and fie=1 how this value came from f=alpha x beta^fie x gamma^fie. Will you please explain or give relevant resoureces. Thank you.😊😊
In EfficientNet, the scaling factors for the depth, width, and resolution are denoted as alpha, beta, and gamma. These scaling factors are typically determined through a systematic search on a predefined grid to find the optimal trade-off between accuracy and computational efficiency.
Hi. actually I don't get that part with feature maps, are you referring to channels by feature maps? If yes, then I don't think this is the correct representation for that? (Which you've drawn like bounding boxes)
Hello, Yes I am referring to channels by feature maps. And I understand what do you mean by correct representation : Feature maps should be stacked one after the other. But why I have choose this image, because my intent is to show that different feature maps carry different part of information from image. And if we use more number of feature maps then we can collect more features from image . But again if we choose lots of feature map it will will work upto some extent but after that there is no use of those extra feature maps as those extra feature maps will degrade the training performance. I hope I made my point :)
Reduce the size of your own window (at the bottom left), will make the PowerPoint presentation more visible. Otherwise its super annoying when your window is cutting off the text and you have not shared any other document as well. Nobody is here to see some enlarged version of the white space in your wall.
This is the WRONG way of giving feedback. Alternatively, you could have said: Great video! I appreciate the content you have provided. However, I suggest that you reduce the size of your video inset a bit so that it doesn't overlap with the powerpoint presentation, as it sometimes hides the text behind it. Additionally, it would be helpful if you share some documents (like the presentation you used in this video) so that we can refer to the unclear parts. Thanks
This is one of the clearest explanations I have seen. You are explaining everything in detail. Thank you!
Glad it was helpful!
It helps me even in 2023 . Hats off to you
Glad it helped 🙂
salutation from malaysia. BEST EfficientNet explanation ever!! TQ
Thank you!
It was straightforward and an amazing way of explaining the point.
Glad it was helpful!
This is very helpful . Thank you!
Glad it was helpful!
Clear and non complicated. Thank you.
Glad it helped!
Thank you for this Well explanation of Compound scaling EfficientNet B0👍
Glad it was helpful!
Easiest explanation ever . Thank you mam.
Welcome
You are explaining like a pro! Thanks mam!
My pleasure 😊
wonderfully explained madam. I really liked this video.
Thank you so much 🙂
Very nice & clear video ma’am. Please keep posting👏👏👏
Sure 😊
Ok you’ve convinced me EffientNet is the one for me
👍
Wow
Ma'am amazing explanation with deap knowledge
No word to say !
Glad you liked it
A great explanation by you, Thank You
Glad my video is helpful!
this video helped me a lot. thank you Aarohi
Glad it helped you :)
Very good and in detailed explanation. Highly recommended
Glad it was helpful!
Great explanation. Thank you.
Glad it was helpful!
You deserve more views and subscriptions! Thanks a lot for this wonderfull lecture
Glad my video is helpful 😊
Thank u very much for this explanations
Happy to help
Amazing! I appreciate you, thank you sooo much
Glad it was helpful!
I can't explain in my word. Really mam you are best teacher😊 on the youtub. But i don't know why your subscribers is very less on TH-cam.😔😔, I will share you channel with my friends.👍😊
That's very kind of you 😊
Great explanation! keep the hard work! Thank you.
Glad it was helpful!
Thank you! Helped me a lot!
Glad it helped!
Thank You Ma'am
Most welcome 😊
Nice explanation mam ❤
Glad you liked it
Thank you mam, very clear and great explanation.
You are welcome 😊
Mam when will you post video for grid search.
Thank you mam. Superb explanation
You are welcome 😊
Your videos are very informative but I have suggestion for you that your picture block covers half of ppt, can you reduce your picture block or you can use green screen. Thanks
Thankyou for this suggestion .. Will implement from next video
Thanks so much for the clear explanation. Could you also please explain Dual Attention Network for segmentation
Will try to cover in my upcoming videos
Thank you mam for teaching us sooo nicely.. I totally agree with @shahidulislamzahid... mam you are too good.
Glad my videos are helping you. keep learning :)
Thankyou Aarohi di 😊
Welcome :)
you have explained it very clearly. can we add skip connection in efficientNet ?
Yes you can modify the Network.
Thank you. Very helpful
Glad it was helpful!
It is really good explanation and helpful enough. When implementation of this will be uploaded? Please make a video of DenseNet.
Implementation of EfficientNet will be available in next 2 days. Will do video on DenseNet after finishing my pipelined videos. Keep Watching 😊
Good mam🎉🎉
Thanks a lot
Very useful, thanks a lot!
Glad it was helpful!
great.tnx
welcome
Hi,
Thanks for the amazing explanation.
I have a question and would request your answer:
If alpha beta and gamma are fixed. φ is obtained using Grid search. Here I want to understand, that during grid search we have to consider different values of φ.
Also, I want to understand what is f here. How does the value of f contribute to the Neural Network?
Please Advise.
Thank You
Sumit
at 32:36 you said the authors already fixed the values of depth width resolution factors, and found phi value, but in that research paper they kept the value of phi as 1 and did a grid search for alpha beta gamma after they found the values they adjusted the phi to upscale different Efficientnets, correct if i am wrong it will be helpful? and the video is good
Brilliant
Thank you!
Thanks a lot!
You're welcome!
Please make a video on 1-D CNN
Sure, will do soon
Maam do you have accumulative explanation on all the CNN models so that a person can get idea of which CNN model has research gap and needs to be worked on it compare to other cnn models on PhD level research gap.
No
Wow, You are awesome. Great explanation. I didn´t understand the last part. How the grid search to obtain (phi) is done?
Will try to cover the grid search in separate video.
Grid search is Hyperparameter tuning technique, researcher would have have done investigation here by trying multiple values and finalized the best or optimized values as depth=1.20;width=1.10 and resolution=1.15
Thanks for explanation. I have question. Can we add this ECA to YOLOv5(put somewhere inside YOLO architecture). Can it decrease Loss Function? Guess)
Mam can u update your playlists by adding the recent videos??so that it will be easy for us to follow.Thank you
Sure I will update today
Hello,
I am using pre trained EfficientNetB0 with ImageNet, I have costume 166 image set, I am using regularizations(drop out and batch normalization )/Augmentation technique. But Getting pretty bad result my val_loss increases and val_accuracy remain constant. could you please help.
Moye moye
Is width scaling means increasing the number of kernels/filters ?
yes, correct
What is mb6 layer
Mam how to calculate phi?
Hello Ma'am,
Can we perform Detection of Malware in windows system (output expected is only in YES/NO) using EfficientNet, or is it only meant for image classification?
EfficientNet is used Image Classification
is it mandatory to use 600*600 dimensions for EfficientNet B7, or can we use smaller dimensions, like say 120*120?
The logic behind efficientnet is to work for high resolution images. If you want to wok for small image size you can use efficientnet B0
@@CodeWithAarohi Okay, got it! Thank you :)
Mam, alpha=1.2,beta=1.1& gamma=1.15 and fie=1 how this value came from f=alpha x beta^fie x gamma^fie. Will you please explain or give relevant resoureces. Thank you.😊😊
In EfficientNet, the scaling factors for the depth, width, and resolution are denoted as alpha, beta, and gamma. These scaling factors are typically determined through a systematic search on a predefined grid to find the optimal trade-off between accuracy and computational efficiency.
Hi. actually I don't get that part with feature maps, are you referring to channels by feature maps? If yes, then I don't think this is the correct representation for that? (Which you've drawn like bounding boxes)
Hello, Yes I am referring to channels by feature maps. And I understand what do you mean by correct representation : Feature maps should be stacked one after the other. But why I have choose this image, because my intent is to show that different feature maps carry different part of information from image. And if we use more number of feature maps then we can collect more features from image . But again if we choose lots of feature map it will will work upto some extent but after that there is no use of those extra feature maps as those extra feature maps will degrade the training performance. I hope I made my point :)
@@CodeWithAarohi i think you are referring to increase number of filters in conv layers
❤❤
Reduce the size of your own window (at the bottom left), will make the PowerPoint presentation more visible. Otherwise its super annoying when your window is cutting off the text and you have not shared any other document as well. Nobody is here to see some enlarged version of the white space in your wall.
Thankyou for the feedback.
This is the WRONG way of giving feedback.
Alternatively, you could have said:
Great video! I appreciate the content you have provided. However, I suggest that you reduce the size of your video inset a bit so that it doesn't overlap with the powerpoint presentation, as it sometimes hides the text behind it. Additionally, it would be helpful if you share some documents (like the presentation you used in this video) so that we can refer to the unclear parts.
Thanks
Hi i am doing research in resnet in thermal image for face recognition , can you help me?
Hello, mail me your requirement at aarohisingla1987@gmail.com Will see if I can help you
Mam, with all due respect. Please don't repeat too much !!!
Thankyou for the feedback and Yes I am already working on this.
Thank u so much
Most welcome 😊
This is one of the clearest explanations I have seen. You are explaining everything in detail. Thank you!
Glad it was helpful!