Open Source Face Analysis with Python

Поделиться
HTML-код
  • Опубликовано: 22 май 2024
  • In this tutorial we go over how to use DeepFace. Deepface is a lightweight face recognition and facial attribute analysis (age, gender, emotion and race) framework for python. Hope you learn something new!
    Timeline:
    00:00 Intro
    00:45 Setup
    03:21 Face Detection
    06:18 Face Verification
    10:11 Face Recognition
    11:39 Facial Attribute Analysis
    12:24 Streaming API
    Follow me on twitch for live coding streams: / medallionstallion_
    My other videos:
    Speed Up Your Pandas Code: • Make Your Pandas Code ...
    Intro to Pandas video: • A Gentle Introduction ...
    Exploratory Data Analysis Video: • Exploratory Data Analy...
    Working with Audio data in Python: • Audio Data Processing ...
    Efficient Pandas Dataframes: • Speed Up Your Pandas D...
    * RUclips: youtube.com/@robmulla?sub_con...
    * Discord: / discord
    * Twitch: / medallionstallion_
    * Twitter: / rob_mulla
    * Kaggle: www.kaggle.com/robikscube
    #python #datascience #facedetection

Комментарии • 70

  • @sefiks
    @sefiks Год назад +111

    Hey 👋 I'm the creator of DeepFace. Thank you for sharing this to the community 🙏

    • @EssaAlshammri
      @EssaAlshammri Год назад +4

      thank you for putting all of those together :)

    • @mignacio79
      @mignacio79 Год назад

      Maybe my question is very naive, but is it possible to train this model to use this tool with the find option to detect animal species? Either in an online camera or from drones or camera traps? Thank you very much for this development.

    • @rex-up9ln
      @rex-up9ln 5 месяцев назад

      Managed to bypass using a 2d photo. Its extremely finicky

  • @Jipzorowns
    @Jipzorowns Год назад +3

    I don't comment much on your video's and in general. Though, I've watched many of your video's and streams and learned so darn much from you. Thank you so much, Rob! Wishing you and your channel all the best in 2023. Again, thanks!

    • @robmulla
      @robmulla  Год назад +2

      Thanks so much for commenting! Its great to know there are people like you out there learning from these videos. It encourages me to keep making them.

  • @niealex4299
    @niealex4299 9 месяцев назад +1

    look forward to more content.. love all the content in your channel.. really praticale in a way demotrate the power of AI/ python... and great eye opener as what AI is capable of

  • @tsaashik8048
    @tsaashik8048 10 месяцев назад

    great job keep grinding👌👌

  • @Infinitay
    @Infinitay Год назад +2

    I’ve been following DeepFace for a while. Thanks for covering it. I’ve been interested in FD and FR and DeepFace made using SOTA models much easier.

    • @robmulla
      @robmulla  Год назад

      Glad it was helpful! Having a single package that works with all the SOTA model is super handy.

  • @ruanwannenburg5471
    @ruanwannenburg5471 Год назад +1

    Great video. Thanks Rob!

    • @robmulla
      @robmulla  Год назад

      Thanks for watching Ruan, hope you learned something new.

  • @dataflex4440
    @dataflex4440 Год назад +1

    Best Channel for data Science Period!.

    • @robmulla
      @robmulla  Год назад

      I appreciate that you think so!

  • @textpower3050
    @textpower3050 Год назад +1

    Thanks for this video. I was confused by all the different models but after watching this video I have a much better understanding now. Is it possible to download the whole final script? I would love to try it out on my machine. Thanks again you are the best.

  • @TheShox79
    @TheShox79 Год назад +1

    Well done! Thank you!

    • @robmulla
      @robmulla  Год назад

      Glad it was helpful!

  • @henkhbit5748
    @henkhbit5748 Год назад +1

    Great, thanks for sharing Rob. For similarity detection, did u try lowering the threshold or use a different function instead of the cosine function for comparing with and without glasses?

    • @robmulla
      @robmulla  Год назад

      Thanks for watching Henk. I didn't try much other than using the defaults but those are definately parameters that you can try and adjusting to see if the results are better for you.

  • @xpinchx
    @xpinchx Год назад +1

    Do you think you'll ever do an MBA? (Market basket analysis) e.g. what e-commerce stores use to group "other people also bought..." lists. I've followed along with a few other text-based tutorials and haven't been able to implement a solution using my own dataset.

    • @robmulla
      @robmulla  Год назад +2

      Great question. I’m not familiar with that so I doubt it. But I might look into it because it sounds really interesting

    • @xpinchx
      @xpinchx Год назад

      @@robmulla you're the best, thanks for considering!

  • @PapiJack
    @PapiJack 5 месяцев назад

    Does anyone know what loss functions these backends are using. Or could you point me to them? I'm new to computeer vision. Thanks.

  • @phattruong7472
    @phattruong7472 Год назад +1

    Could you make video review the laptop you're using and give us any advice on choosing the laptop for data science? Thanks in advance.

    • @robmulla
      @robmulla  Год назад

      My laptop is nothing special actually, but I really like my desktop that I built myself with custom parts.

  • @phattruong7472
    @phattruong7472 Год назад +2

    Hello Rob Mulla, I wonder if you could share some courses data beginner should take, especially about statistics, maths, and machine learning courses. Thanks a lot.

    • @robmulla
      @robmulla  Год назад

      There are a lot out there. Check udamy or edx. Or just watch RUclips and streams!

  • @jdmeyer05
    @jdmeyer05 Год назад

    Where did you get that T-shirt?

  • @GPHReviews
    @GPHReviews Год назад +1

    Hey rob, I am getting the type as list for result 12:20, but as u r using keys to index, how u got the dict in return from deep face ? Thankyou

    • @robmulla
      @robmulla  Год назад

      Thanks for the feedback. It's been some time since I created the video. It's possible the api has changed. You can check the github repo to see if they've made major changes and let me know. Hope that helps.

  • @muskanadwani8111
    @muskanadwani8111 Год назад

    I am getting output of facial analysis as type "list" instead of "dictionary"
    pls help me

  • @resresres1
    @resresres1 Год назад +1

    I'd like to request a video. Can you make a video on reducing .exe file size (either with one file or one directory) when converting .py to .exe? With it being more in depth than simply a generic, "use a clean virtual environment and pyinstaller" and with using some of the popular libraries like pandas/numpy/etc. It's honestly extremely difficult to get any good information regarding this and there are not really any good tutorials on accomplishing this on how to create the smallest exe.

    • @robmulla
      @robmulla  Год назад

      Hmm. Good suggestion. I don’t know much about exe files but I can try to learn and will let you know.

  • @irfanjaafar3570
    @irfanjaafar3570 Год назад +1

    VGG-Face is pretty slow, I would recoment using SFace if anyone want a faster recognition. Great video

    • @robmulla
      @robmulla  Год назад

      Hey thanks for the comment. I’ll check VGG Face out.

  • @rathgrams
    @rathgrams 4 месяца назад

    Do data scientists only use ipynb and not py files?

  • @tatelangdon4809
    @tatelangdon4809 4 месяца назад

    Hi, I followed ur tutorial to instal and use deepfake, everything else worked perfectly, however for the streaming about, it just outputs db path not detected, am not a professional, can anyone just explain what does that mean, and like how to I fix it?? thanks!!

  • @bagusandrewijaya
    @bagusandrewijaya 6 месяцев назад

    how to make deepface.find result as json response?

  • @user-ld5dn3fv4m
    @user-ld5dn3fv4m Год назад +1

    Hello. You don't have video streaming data. I think it will be good if you shoot a video on the topic of streaming data in pandas with a binance stream. Next, we use any model on streaming data.

    • @robmulla
      @robmulla  Год назад +1

      Thanks for the suggestion. That's a good idea. I don't know much about streaming data yet, but I think it could potentially make a really interesting video.

  • @SamsungJ-hi5gb
    @SamsungJ-hi5gb Год назад +1

    Can you make video tutorial about implementation of DeepFace using Raspberry Pi for Door Lock? Thanks Mr.Rob.

    • @SamsungJ-hi5gb
      @SamsungJ-hi5gb Год назад +1

      I'm from indonesia

    • @robmulla
      @robmulla  Год назад

      Hey! That’s a very specific request. I can’t guarantee I will make that video but hopefully my videos will help you get started!

  • @r3yjt
    @r3yjt 10 месяцев назад

    Is it also for Mac?

  • @syahrulmaulanaw6965
    @syahrulmaulanaw6965 Год назад +1

    Hi Rob, do you think people can land a job as a data analyst or data scientist without data science degree?
    I'm currently pursuing my Electrical Engineering degree (Control system eng) but got interested in data science field :)

    • @robmulla
      @robmulla  Год назад +1

      Great question. I think that working to build up a resume showing your skills in data science is important. Kaggle, projects on github, etc. But also if you are in school try seeing if minoring in computer science is an option. IMO if you have a technical degree like EE it will be helpful to land a job even if it's not specifically in CS/data science. Just my experience though.

    • @syahrulmaulanaw6965
      @syahrulmaulanaw6965 Год назад

      @@robmulla thanks for the advice Rob. Actually my degree itself is close to CS in some topics (since i take the concentration in CSE), thats why I have a thought to take the data science more serious. But, Im gonna think about what you've said, cheers

  • @us30legendfx15
    @us30legendfx15 Год назад +1

    how can i import the pictures using googlecollab ?

    • @robmulla
      @robmulla  Год назад

      You could upload them or try linking to your google drive?

  • @neeraj.kumar.1
    @neeraj.kumar.1 Год назад +2

    Please make a video on ANPR using Yolo.

    • @robmulla
      @robmulla  Год назад +1

      Thanks for posting! I might do something like that but I think there are already some pretty good videos on youtube about it.

    • @neeraj.kumar.1
      @neeraj.kumar.1 Год назад +2

      @@robmulla
      I think you can explain it in more better way.

  • @sidarth404
    @sidarth404 Год назад +1

    Can you pls make video on " speech to text and back to speech instantly" with gtts free neural voice.

    • @robmulla
      @robmulla  Год назад

      Oh, that would be a good video. I have one on speed to text using OpenAI whisper: ruclips.net/video/HbY51mVKrcE/видео.html
      I don't have anything on text to speech yet though. Maybe in the future.

  • @raczyk
    @raczyk Год назад

    Can DeepFace do race recognition?

  • @zac_vaughn
    @zac_vaughn Месяц назад

    To be fair... You do genuinely look like two different people in those images.

  • @macemastalerz
    @macemastalerz Год назад +1

    How can we give this information to a conversational ai?

    • @robmulla
      @robmulla  Год назад

      You can do anything you want with the outputs, but I'm not sure the best way to feed it to an AI would be. Copy/pase the results :D ChatGPT is only accessable via the browser right now.

  • @user-in7bf1fq4r
    @user-in7bf1fq4r Год назад +1

    great video! Can you please share the code?

    • @robmulla
      @robmulla  Год назад

      Hey! I’ll try to find it and post. Thanks for watching.

  • @ThatWeirdKiidOnYT
    @ThatWeirdKiidOnYT 5 месяцев назад

    The key to defeating the state sponsored state of the art facial tracking is..Glasses

  • @RAZZKIRAN
    @RAZZKIRAN Год назад +1

    please share the code

    • @robmulla
      @robmulla  Год назад +1

      I believe most of the code was taken directly from the DeepFace docs.

  • @CaribouDataScience
    @CaribouDataScience Год назад +1

    Wow! I never would guess that you were of Asian decent?

    • @robmulla
      @robmulla  Год назад +2

      Haha. I'm half Indian so technically it's not wrong.

  • @benkhemismarwen2431
    @benkhemismarwen2431 Год назад +1

    its unlucky that the accuracy of face recognition is nothing near what big company achieved.

    • @robmulla
      @robmulla  Год назад

      Yes, but they can be improved!