DCGAN | Deep Convolutional Generative Adversarial Network

Поділитися
Вставка
  • Опубліковано 10 січ 2025

КОМЕНТАРІ • 60

  • @pramodhbr4190
    @pramodhbr4190 3 роки тому +2

    Awesome mam please keep continuing making good content like this

  • @Shashank-vv4ex
    @Shashank-vv4ex 2 роки тому

    Thankyou Maam for an amazing explanation and implementation.

  • @jackhardy813
    @jackhardy813 3 роки тому +2

    as always great content!

  • @aiforeveryone
    @aiforeveryone Рік тому

    Top notch explaination

  • @ambujpawar
    @ambujpawar 3 роки тому +5

    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!

    • @CodeWithAarohi
      @CodeWithAarohi  3 роки тому +1

      Glad my content is helpful and Sure I will manage my videos playlist wise soon.

  • @mimran-eo6wp
    @mimran-eo6wp 3 роки тому

    Thanks, mam appreciable effort

  • @mohammadwahba3077
    @mohammadwahba3077 3 роки тому

    Excellent - GOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOd

  • @turalkerimov4022
    @turalkerimov4022 2 роки тому

    You are BEST!

  • @KasimKhan-lk4sw
    @KasimKhan-lk4sw 3 роки тому

    fabulous

  • @faizack
    @faizack Рік тому

    Awesome

  • @rahiegadekar3422
    @rahiegadekar3422 3 роки тому

    That was really helpful ! Please do make a video on SRGAN.

  • @joafin19
    @joafin19 2 роки тому +1

    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.

    • @Rakesh_Seerla
      @Rakesh_Seerla Місяць тому

      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

  • @subashchandrapakhrin3537
    @subashchandrapakhrin3537 3 роки тому +2

    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.

    • @CodeWithAarohi
      @CodeWithAarohi  3 роки тому

      Hi, sorry I dont have video for this right now.

    • @usmanjalil6711
      @usmanjalil6711 Рік тому

      see concept of functional api for deep learning models you will get an idea

  • @sandhyapotadar358
    @sandhyapotadar358 Рік тому

    Mam can we have kernal size is usually 3by3 odd filter

  • @Athulyanklife
    @Athulyanklife 2 роки тому

    How to try DC Gan to increase image dataset size on custom dataset ?

  • @soravsingla8782
    @soravsingla8782 10 місяців тому

    Perfect

  • @SanjaySingh-qf4tk
    @SanjaySingh-qf4tk 3 роки тому

    mam how to impliment speech enhancement using GAN network. pls guide me.

  • @drhope1234
    @drhope1234 3 роки тому

    thank you it was interested ^^

  • @ravilourembam9412
    @ravilourembam9412 3 роки тому

    good video maam... please can u do a tutorial on cnn+elm hybrid?

  • @cool_videos6016
    @cool_videos6016 3 роки тому

    It's very good mam can you please make a video on anomaly detection in videos

  • @DilawarShah-g9f
    @DilawarShah-g9f Рік тому

    Dear Aarohi, i want to augment MIAS dataset. kindly guide me

    • @CodeWithAarohi
      @CodeWithAarohi  Рік тому +1

      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.

    • @DilawarShah-g9f
      @DilawarShah-g9f Рік тому

      @@CodeWithAarohi madam the images generated by our GAN are not good from MIAS. What we should do?

  • @terryterry3733
    @terryterry3733 3 роки тому

    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?

    • @CodeWithAarohi
      @CodeWithAarohi  3 роки тому +1

      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

  • @siddharthtyagi1254
    @siddharthtyagi1254 3 роки тому

    Is there any difference b/w cycle gan and dcgan. Great video.

    • @CodeWithAarohi
      @CodeWithAarohi  3 роки тому

      Cycle GAN perform image to image translation rather than image synthesis from a random vector.

    • @siddharthtyagi1254
      @siddharthtyagi1254 3 роки тому

      Mam please make a video on cycle gan also

    • @CodeWithAarohi
      @CodeWithAarohi  3 роки тому

      @@siddharthtyagi1254 Sure

  • @rajshreesrivastava7650
    @rajshreesrivastava7650 2 роки тому

    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

    • @CodeWithAarohi
      @CodeWithAarohi  2 роки тому

      you can use generator to generate fake samples and then you can use that data as an input to your any classification model.

  • @ykakde
    @ykakde 6 місяців тому

    😍😍😍😍

  • @roohiali5713
    @roohiali5713 3 роки тому

    mam please make video on GAN using MATLAB..

    • @CodeWithAarohi
      @CodeWithAarohi  3 роки тому

      Hi Roohi... I work using Python but will try in future to work using Matlab

  • @Rakesh_Seerla
    @Rakesh_Seerla Місяць тому

    Those generated images looks like scary 😂

    • @CodeWithAarohi
      @CodeWithAarohi  Місяць тому

      Haha, I know, right? 😂 It happens sometimes when the model gets creative! Thanks for checking out the video!

  • @vishalomance
    @vishalomance 2 роки тому

    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

  • @chronologyofai672
    @chronologyofai672 3 роки тому

    Mam please cyclegan and progan and also two video music generation and super resolution please

  • @urvashi669
    @urvashi669 2 роки тому

    Mam can you please make a video for cartoonGAN?

  • @manishreddy9799
    @manishreddy9799 Рік тому

    boring