Fine-tune ConvNeXT for image classification [end-to-end tutorial]

Sdílet
Vložit
  • čas přidán 13. 09. 2024
  • Hi guys!
    In this video, I show how to fine-tune ConvNeXT, a state-of-the-art model by Facebook AI for image classification on a custom dataset. I leverage the new 🤗 ImageFolder feature, and train the model in native PyTorch.
    Timestamps:
    0:55 Intro to ConvNeXT
    1:35 Loading a custom dataset using ImageFolder
    6:40 Push dataset to the 🤗 hub
    8:57 Process dataset
    18:10 Define model
    21:20 Prepare training of the model
    34:30 Start training!
    36:46 Push model to the hub
    38:15 Performing inference (with widget)
    44:20 Pipeline API
    Colab notebook: github.com/Nie...

Komentáře • 22

  • @robosergTV
    @robosergTV Před rokem

    awesome, thanks Niels!

  • @satpalsinghrathore2665
    @satpalsinghrathore2665 Před 2 lety +1

    Would love to see more videos!

  • @pillaideepakb
    @pillaideepakb Před 11 měsíci

    Great video. Thank you Niels

  • @fthdev1
    @fthdev1 Před 4 měsíci

    Thanks 😎

  • @MerveNoyan
    @MerveNoyan Před 2 lety

    Thanks for the great tutorial!

  • @nandinisarker6123
    @nandinisarker6123 Před rokem

    Hi Niels, this video helped me a lot in my master's degree study. I have been trying to implement an idea to use PSO in a pre-trained model as you showed, but not sure how to start. Could you please make a guideline video?

  • @omarboukadoum8294
    @omarboukadoum8294 Před 6 měsíci

    Thank you for clarifying this course. Could you please share the code project related to ConvNext? I appreciate it greatly.

  • @nandinisarker6123
    @nandinisarker6123 Před 2 lety

    Very nice tutorial on image classification, thank you for your contribution in huggingface team. Can you please show how to plot accuracy matric while using pre-trained model like this?

  • @LoretoParisiLabs
    @LoretoParisiLabs Před 2 lety

    Well done!

  • @krishnabhagavan9621
    @krishnabhagavan9621 Před 2 lety

    Great job buddy

  • @martinsfolefac1820
    @martinsfolefac1820 Před 2 lety

    Great job

  • @nikdab
    @nikdab Před 2 lety +1

    Great work. Can you please upload the colab notebook? Also, is there a way of uploading data from local device?

    • @NielsRogge
      @NielsRogge  Před 2 lety +1

      Hi, you can upload from local folder(s) by providing the data_dir argument instead of data_files when using ImageFolder. Details here: huggingface.co/docs/datasets/image_process

  • @syedmujtabahassanrizvi2129

    How can we find validation accuracy, test accuracy, and f1 scores?

  • @ronancadmielcastro2200
    @ronancadmielcastro2200 Před 6 měsíci

    Hello, thank you so much for the tutorial. Just a question: can ConvNeXt be used in MATLAB? And if yes, any ideas on these?

    • @NielsRogge
      @NielsRogge  Před 6 měsíci

      Hi, Matlab isn't really used in AI, everything is in Python

    • @ronancadmielcastro2200
      @ronancadmielcastro2200 Před 6 měsíci

      Thank you for the response. Can I ask if for you, what are the most cutting edge state of the art image classifier in 2024? thank you very much, this is very helpful for my PhD dissertation.

  • @syedmujtabahassanrizvi2129

    please upload the colab notebook on git and share the link in description or in comment section. secondly How can we find validation accuracy, test accuracy, and f1 scores?

  • @bousejin
    @bousejin Před 2 lety

    Thank you so much! Zoom a little bit into your code for your next videos, please!

  • @SyedMujtabaHassanRizvi

    can you please share this notebook with me? How can we find validation accuracy, test accuracy, and f1 scores?

    • @NielsRogge
      @NielsRogge  Před 2 lety

      Hi, I've added a link to the notebook at the end of the description.