Hi, thanks for the super great content on your channel. I see Is it possible for you to organize playlists on your channel in a way like "Computer Vision", 'NLP", etc? Thanks again for the content!
Could you explain why you chose 3 convolution layer of 128 256 512 and kernel size 4 and stride 2 for generator and 64 128 128 for discriminator? Tank you.
We do increase filters upto ur computation not only 512 you can do 1024.. and more there is no problem finally you can convert them into original shape that's it Stride 2,2... Faster conversions of our sizes like 32*32*256 to 64*64*512 both modified here grid and filters Kernel size 4 based on our require ment 3 or 4 especially for edge detection of our image In generator we do grid decrease and filters increase for less computation on computer that's it as like CNN
Dear Aarohi do you have any videos on Multi-Input Models although I know it and learnt it from page 238 - 240 of Deep Learning with Python by François Chollet but your video will be more helpful Thank You.
Prepare the dataset: Download the MIAS dataset and preprocess the images to a suitable format. You can resize the images and convert them to grayscale if necessary. Train a DCGAN model: Use a deep convolutional generative adversarial network (DCGAN) to generate new images. DCGANs are a type of neural network that can generate new images by learning the distribution of the input images. You can train the DCGAN model on the MIAS dataset to generate new mammogram images. Augment the dataset: Once you have trained the DCGAN model, you can use it to generate new images that can be added to the original MIAS dataset. You can also use data augmentation techniques such as rotation, scaling, and flipping to increase the size of the dataset further.
Hi Aarohi in ( L-10 Image Classification using Perceptron | Singla Layer Neuron ) u said we have to 1) resize all the images 2)convert images into array. but in this DCGAN session u didnt do it . can u pls tell me it is not required in DCGAN?
Hello Terry, In DCGAN also we are resizilng the images. Check cell no 15 of DCGAN Code (github.com/AarohiSingla/DCGAN-Generating-Fake-Faces-with-DCGAN/blob/main/dcgan.ipynb) here image_size=(64, 64) is the size to resize all the images present in that folder
Hello mam, Nice explanation sir Can u plZ tell how we can use gan for data augmentation and deep learning alexnet /resnet-50/ vgg 16 for classification
Think of the DCGAN using a 7x7 mask for the celebA dataset. a. How would you set up the problem to select the activation functions and optimizersfor this using a genetic algorithm? b. Describe how the computation speed of this model can be improved, or if a DCGANis the appropriate method to use. c. What is an alternate model that can run faster with better results? Madam can you please explain this to me I hope you will Thank you
Awesome mam please keep continuing making good content like this
Thankyou Maam for an amazing explanation and implementation.
My pleasure 😊
as always great content!
Glad you think so!
Top notch explaination
Thank you!
Hi, thanks for the super great content on your channel. I see Is it possible for you to organize playlists on your channel in a way like "Computer Vision", 'NLP", etc? Thanks again for the content!
Glad my content is helpful and Sure I will manage my videos playlist wise soon.
Thanks, mam appreciable effort
Thanks for liking
Excellent - GOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOd
Thanks for listening
You are BEST!
Glad my video is helpful
fabulous
Thanks
Awesome
Thanks!
That was really helpful ! Please do make a video on SRGAN.
Sure
Could you explain why you chose 3 convolution layer of 128 256 512 and kernel size 4 and stride 2 for generator and 64 128 128 for discriminator? Tank you.
We do increase filters upto ur computation not only 512 you can do 1024.. and more there is no problem finally you can convert them into original shape that's it
Stride 2,2... Faster conversions of our sizes like 32*32*256 to 64*64*512 both modified here grid and filters
Kernel size 4 based on our require ment 3 or 4 especially for edge detection of our image
In generator we do grid decrease and filters increase for less computation on computer that's it as like CNN
Dear Aarohi do you have any videos on Multi-Input Models although I know it and learnt it from page 238 - 240 of Deep Learning with Python by François Chollet but your video will be more helpful Thank You.
Hi, sorry I dont have video for this right now.
see concept of functional api for deep learning models you will get an idea
Mam can we have kernal size is usually 3by3 odd filter
How to try DC Gan to increase image dataset size on custom dataset ?
Perfect
Thanks!
mam how to impliment speech enhancement using GAN network. pls guide me.
thank you it was interested ^^
Glad to know
good video maam... please can u do a tutorial on cnn+elm hybrid?
It's very good mam can you please make a video on anomaly detection in videos
Sure will do soon
Dear Aarohi, i want to augment MIAS dataset. kindly guide me
Prepare the dataset: Download the MIAS dataset and preprocess the images to a suitable format. You can resize the images and convert them to grayscale if necessary.
Train a DCGAN model: Use a deep convolutional generative adversarial network (DCGAN) to generate new images. DCGANs are a type of neural network that can generate new images by learning the distribution of the input images. You can train the DCGAN model on the MIAS dataset to generate new mammogram images.
Augment the dataset: Once you have trained the DCGAN model, you can use it to generate new images that can be added to the original MIAS dataset. You can also use data augmentation techniques such as rotation, scaling, and flipping to increase the size of the dataset further.
@@CodeWithAarohi madam the images generated by our GAN are not good from MIAS. What we should do?
Hi Aarohi in ( L-10 Image Classification using Perceptron | Singla Layer Neuron ) u said we have to 1) resize all the images
2)convert images into array. but in this DCGAN session u didnt do it . can u pls tell me it is not required in DCGAN?
Hello Terry, In DCGAN also we are resizilng the images. Check cell no 15 of DCGAN Code (github.com/AarohiSingla/DCGAN-Generating-Fake-Faces-with-DCGAN/blob/main/dcgan.ipynb) here image_size=(64, 64) is the size to resize all the images present in that folder
Is there any difference b/w cycle gan and dcgan. Great video.
Cycle GAN perform image to image translation rather than image synthesis from a random vector.
Mam please make a video on cycle gan also
@@siddharthtyagi1254 Sure
Hello mam,
Nice explanation sir
Can u plZ tell how we can use gan for data augmentation and deep learning alexnet /resnet-50/ vgg 16 for classification
you can use generator to generate fake samples and then you can use that data as an input to your any classification model.
😍😍😍😍
mam please make video on GAN using MATLAB..
Hi Roohi... I work using Python but will try in future to work using Matlab
Those generated images looks like scary 😂
Haha, I know, right? 😂 It happens sometimes when the model gets creative! Thanks for checking out the video!
Think of the DCGAN using a 7x7 mask for the celebA dataset.
a. How would you set up the problem to select the activation functions and optimizersfor this using a genetic algorithm?
b. Describe how the computation speed of this model can be improved, or if a DCGANis the appropriate method to use.
c. What is an alternate model that can run faster with better results?
Madam can you please explain this to me
I hope you will
Thank you
Mam please cyclegan and progan and also two video music generation and super resolution please
Will try to cover soon
Mam can you please make a video for cartoonGAN?
Sure
boring
Thank you for the feedback.