AI on the Jetson Nano LESSON 39: Face Recognition and Identification with OpenCV

Поделиться
HTML-код
  • Опубликовано: 3 фев 2025

Комментарии • 70

  • @theteenengineer7589
    @theteenengineer7589 4 года назад +17

    mr Paul is a national treasure :)

    • @paulmcwhorter
      @paulmcwhorter  4 года назад

      Very kind of you!

    • @modraccin9514
      @modraccin9514 4 года назад

      A gobal treasure! Thanks for your inclusive thinking, Paul!

  • @jeffhuevel2517
    @jeffhuevel2517 4 года назад +1

    Paul, The work that you do to keep this project moving is amazing. The community is very grateful !

  • @SH-vv7zq
    @SH-vv7zq 4 года назад +1

    Great lesson Paul. Worked first time. Well explained, and still love the fact you get live errors as you code on-the-fly; brings a sense of realism to everyday coders. Keep up the brilliant tutorials.

  • @SuperAlex5
    @SuperAlex5 4 года назад

    Every lessons are incredible. Very practical for all learners. I mean, it seems the best and perfect way to teach.

  • @mangalarupasinghe
    @mangalarupasinghe 4 года назад +1

    Excellent Tutorial Paul, Great teaching with simplified explanations. Thanks !

  • @boxertwo
    @boxertwo 4 года назад

    As of this writing I'm a few lessons behind but I really enjoy the content!! Keep up the great work!! Having a blast going through the lessons:).

  • @Mircea007
    @Mircea007 4 года назад

    Hi Paul, this is getting more exiting by the day!

  • @ricardobjorkeheim775
    @ricardobjorkeheim775 3 года назад

    Great lesson - worked perfectly for me after finding all the errors I made :)!

  • @sanfinity_
    @sanfinity_ 3 года назад

    WOW just like that I can now train a model for my face😃😃 Thank you sir for an another wonderful lesson

  • @geeksatlarge
    @geeksatlarge 4 года назад

    Thanks Paul! This one was a little rough on me, specifically that for loop. I *think* I have it now, but we'll see. Now on to the HW.

  • @danbishop4035
    @danbishop4035 3 года назад

    This is really great stuff I have a feeling that I will need to upgrade to a better Jetson. Excellent video as usual :)

  • @Jason-ot6jv
    @Jason-ot6jv 4 года назад +6

    man this looks so freaking cool. Just wish it wasn't so expensive lol

    • @TheRealFrankWizza
      @TheRealFrankWizza 4 года назад

      I just got the 2 gig. It's much cheaper, and it works just fine.

  • @JSEEngineering
    @JSEEngineering 4 года назад

    Excellent exercise - Great Teacher Paul

  • @CodingScientist
    @CodingScientist 4 года назад +1

    Waiting for the count down :-)

  • @hervecoquet3046
    @hervecoquet3046 4 года назад

    Execelent and crystal clear as usual.

  • @teachusmore
    @teachusmore 4 года назад

    Paul, if you wanted to encode multiple images of the same person in order to improve accuracy would this work? :
    example...
    nancyFace=face_recognition.load_image_file(r"/home/pi/documents/nancy-1.jpeg")
    nancyFace=face_recognition.load_image_file(r"/home/pi/documents/nancy-2.jpeg")
    face_recognition.face_encodings(nancyFace)[0]
    ...or...
    nancyFace=face_recognition.load_image_file(r"/home/pi/documents/nancy-1.jpeg")
    face_recognition.face_encodings(nancyFace)[0]
    nancyFace=face_recognition.load_image_file(r"/home/pi/documents/nancy-2.jpeg")
    face_recognition.face_encodings(nancyFace)[0]
    I'm not getting errors on either method, but wanted to know which is the correct way of doing it and if it's truly encoding multiple photos. Thank you & great channel!

    • @paulmcwhorter
      @paulmcwhorter  4 года назад

      I think the way I would do it would be to put a folder called nancy. Then in the folder a picture 0.jpg, 1.jpg, 2.pjg . . .then train on all the pictures of nancy, but get the ID name not from the file, but from the folder.

    • @teachusmore
      @teachusmore 4 года назад

      @@paulmcwhorter I'll try that, thanks for the help!

    • @teachusmore
      @teachusmore 4 года назад

      Doesn't seem to work, I tried (filepath)/ and I also tried (filepath)/*.jpg .

  • @tomaszb7156
    @tomaszb7156 4 года назад

    It will be interesting to make all this as unsupervised learning and see how jetson nano handle this

  • @AlanHeigl
    @AlanHeigl 4 года назад

    great just great... can't wait for more

  • @HA-wo3we
    @HA-wo3we Год назад

    fantastic vid again. is the model for face recognition model based on tensorflow trained model?

  • @donthomas9669
    @donthomas9669 4 года назад

    Paul, thank you for all you do. I finally got my Nano and am looking forward to starting working with it. I also want to let you know the Jetson Nano tutorials are only partially showing up on your site now. Also, the links in your previous lessons to parts are often linking to discontinued parts or out of stock parts in amazon. Its hard to find the same exact parts you are using now. I know this is in no way your fault, i just wanted to let you know about this.

  • @eranfeit
    @eranfeit 4 года назад

    thanks again for a wonderful lesson

  • @kevinshaver5613
    @kevinshaver5613 4 года назад

    Paul, I was thinking that calling an avi file after identifying a face would be a cool way of making the Nano appear to talk.Outputting this to a small audio amp mounted to the Nano would complete the effect.

    • @somebody9033
      @somebody9033 4 года назад +1

      You can use espeak for that

  • @admizasa9215
    @admizasa9215 4 года назад

    Hi, outstanding project and love the way you've built up to the final reveal. Not watched the previous episodes yet but would like to know if it is possible to extract faces from a live video stream, compare them with a folder of known images and highlight any unknown faces?

  • @JSEEngineering
    @JSEEngineering 4 года назад

    Excellent - Wunderbar!

  • @wanlongpeng3782
    @wanlongpeng3782 3 года назад +2

    i always have the bug:
    Traceback (most recent call last):
    File "/home/nano/Desktop/PYPRO/faceRecognizer/demoImages-master/faceRecognize-2.py", line 23, in
    matches=face_recognition.compare_faces(Encodings,face_encoding)
    File "/usr/local/lib/python3.6/dist-packages/face_recognition/api.py", line 226, in compare_faces
    return list(face_distance(known_face_encodings, face_encoding_to_check)

  • @stuartmallett6334
    @stuartmallett6334 4 года назад

    Great tutorial, thank you.

  • @quaternion-pi
    @quaternion-pi 4 года назад +1

    Impressive! The cv2 module no longer supports autocomplete (intellisense) - any way to restore it? i.e. cv2.[*****] makes no appropriate suggestions. The face_recognition module autocomplete works fine. Great lesson - worked perfectly!

    • @paulmcwhorter
      @paulmcwhorter  4 года назад

      Yes, it broke with jetpack 4.3. I have a solution in an upcoming lesson. I miss it, as I can not remember all the details on order of parameters it wants. Solution coming soon.

    • @KamalSehairi
      @KamalSehairi 4 года назад +1

      I get help from from good man who shared with us this trick and I sharing with you
      to add autocompletion (intellisense) option
      do the following
      CTRL+SHIFT+P type Preferences:Open Settings(JSON) Add these settings inside the { } : "python.linting.pylintArgs": ["--generate-members", "--extension-pkg-whitelist=cv2"], "python.autoComplete.extraPaths": ["/usr/lib/python3.6/dist-packages/cv2/python-3.6"]

    • @everythingrobotics
      @everythingrobotics 3 года назад

      @@KamalSehairi Thank you so much! I had the same problem

  • @raghavneema8849
    @raghavneema8849 4 года назад

    Sir ,thank you very much for sharing such wonderful knowledge. I really love your videos. I have a request that can you please make a new series on arduino with Python 3 on windows 10.
    Sir I really appreciate all your hard work please keep making such awesome videos

    • @paulmcwhorter
      @paulmcwhorter  4 года назад

      The issue with that series is to get the data tranfered from arduino to python. Once you can do that, all the other examples should work. Look at the series on the IMU that I did. In that lesson I showed all the things to get arduino working with python. If you watch that series you will learn what you need to know.

    • @raghavneema8849
      @raghavneema8849 4 года назад

      Thank you so much for the reply you are very good teacher you explain all your knowledge in the most simplest way possible thank you so much for that
      May god bless you
      One day I am going to show up to you’re home in west texas to thank you

  • @supertonyxd
    @supertonyxd 3 года назад

    very nice, thank you!

  • @AppieBing
    @AppieBing 4 года назад

    Paul i did the homework, but i am kinda dissapointed that a few pictures pictures returned 'Unknown Person' including yours with chase and the ronald reagan ones.....Great Lesson as always though :)

  • @HUDSON-EV
    @HUDSON-EV 3 года назад

    Thank you.

  • @kevinyu9934
    @kevinyu9934 4 года назад

    freaking awesome

  • @SteJuMusic
    @SteJuMusic 4 года назад

    You are doing a Fantastic Series here. I enjoy following these steps. I have one question. I installed jtop, you mentioned in the xavier nx introduction. On the nano, with jetpack 3.1, it shows that cuda is not compiled. Does this have a bad influence on the face_recognician performance? Thanks a lot for your Great work.

    • @SteJuMusic
      @SteJuMusic 4 года назад

      Sorry, i have installed jetpak 4.3 - my installation is Equal Touch yours

  • @Jure1234567
    @Jure1234567 4 года назад

    I have just started using jetson nano, trying to install tensorflow from nvidia, but.... The installation freezes, I even have put ice on the heatsink, so it is cool like 20 C, but still it is frozen (( I dunno how much time to wait, scared if something breaks in the sd if I cut off the power

  • @MrKrishnanandaKHegde
    @MrKrishnanandaKHegde 2 года назад

    Is it possible to interface with an IP Camera instead of a USB?

  • @wishicouldarduino8880
    @wishicouldarduino8880 4 года назад

    Running in to or out of some snags with the nano .........not programming that's always in a snag lol but hook up it has a lot of potential but no plug ins no decent pinouts it has the power to run a complete robot that is a real robot design but no real way to hook it all up ! This thing needs to tie into ultrasonic sensors,tf mini.360 lidar which is hooked up to the only remaining USB port interrupt sensors 5 volt relay ,voice control system right now I'm using mp3 players but will integrate e speak soon ,on and on looks like I'm going to have to make many boards to hook it all up I don't want to even talk about i2c connections i2c connections ? I need about 2 million .👍😀 cool video!

  • @SoulfulAish
    @SoulfulAish 2 года назад

    Hello, I did every steps as you did, still I am getting error "No module named face_recognition". Can someone help me how to fix this?

  • @ricardobjorkeheim775
    @ricardobjorkeheim775 3 года назад

    to remove the "canberra error write on terminal: sudo apt-get install libcanberra-gtk-module
    you must place the image on the screen with: cv2.moveWindow("myWindow",0,0) to show the image of trump and nancy, other ways it will just blink and the image will disappear.

  • @wishicouldarduino8880
    @wishicouldarduino8880 4 года назад

    Is there a single piece of code for this ????????? My project is far to large to carry off on side folders and distractions im never going to use if so cool if not I have to wait till absolute last👍😀

    • @paulmcwhorter
      @paulmcwhorter  4 года назад

      I have no idea what you are saying

    • @wishicouldarduino8880
      @wishicouldarduino8880 4 года назад

      @@paulmcwhorter is there like one long sketch where it has everything up to this point I got behind on these and I didn't know if there are a bunch of folders or only one that covers everything ? Reason is for building one node for this in ros I haven't tried yet to scared to mess up on my Fifth flash! Lol 👍😂

    • @wishicouldarduino8880
      @wishicouldarduino8880 4 года назад

      @@paulmcwhorter if I flash one more time ray Stevens is going to write a song about it lol I said Ethel don't look but it was too late. She'd already been flashed!...lol😁

    • @paulmcwhorter
      @paulmcwhorter  4 года назад +1

      No, because I am not here to give you code to copy and paste, I am here to teach you an important topic area. I teach, you learn, then you write your own code.

    • @wishicouldarduino8880
      @wishicouldarduino8880 4 года назад

      @@paulmcwhorter ahhhh I see I never copy and paste just for that reason I at least hand copy I didn't even copy and paste the string though it did take me a while to figure out ('NanoCam') instead of picam but I do have some pride so I always hand copy no matter how large the piece of code that's why I was wondering if there was a obscene long code for the hand copy challenge . I understand though.👍😀

  • @richardw782
    @richardw782 3 года назад

    I named my program differently, but the code is identical. Getting a error:
    s4m@s4m-desktop:~/Desktop/pyPro$ /usr/bin/python3 /home/s4m/Desktop/pyPro/faceRecognizer/faceRecognize-2.py
    4.1.1
    Traceback (most recent call last):
    File "/home/s4m/Desktop/pyPro/faceRecognizer/faceRecognize-2.py", line 23, in
    matches = face_recognition.compare_faces(Encodings, face_encoding)
    File "/usr/local/lib/python3.6/dist-packages/face_recognition/api.py", line 226, in compare_faces
    return list(face_distance(known_face_encodings, face_encoding_to_check)

    • @richardw782
      @richardw782 3 года назад

      Found the problem. Dont make this mistake : donEncode = face_recognition.face_encodings(donFace[0]) should be: donEncode = face_recognition.face_encodings(donFace)[0]

  • @KamalSehairi
    @KamalSehairi 4 года назад +1

    Excellent Tutorial. Thanks a lot
    One objection in this lesson is using politicians photos especially Trump. LOL

    • @paulmcwhorter
      @paulmcwhorter  4 года назад +2

      The purpose of using politicians is that the images are not copyrighted, everyone recognizes them, and there are no issues with trademark. It was not a political statement, and I showed people from both parties. You are the one who made it political. Just so you know, I dont publish comments with political opinion.

    • @KamalSehairi
      @KamalSehairi 4 года назад +1

      Paul McWhorter
      Thank you Professor, I was just kidding, no offense
      I know even datasets may have copyrights and many have been deleted for this purpose
      www.google.com/amp/s/fortune.com/2019/06/07/microsoft-facial-recognition/amp/

  • @TheRealFrankWizza
    @TheRealFrankWizza 4 года назад

    He who denied it, supplied it.

  • @wefow1
    @wefow1 3 года назад

    Haha I made a silly mistake in the code "image=face_recognition.load_image_file(........)"
    I added a negative sign in front of the code "face_recognition.load_image_file...." and got a very scary picture
    That's why my program did not recognize two faces

  • @matthewjohnson2497
    @matthewjohnson2497 4 года назад

    Aagghh! Code-OSS, has just disappeared, what's happened?

  • @OZtwo
    @OZtwo 3 года назад

    wow, my B01 seemed to be much faster. Only took 10 seconds?

  • @kenstephens221
    @kenstephens221 4 года назад

    Got the homework done. It named all participants correctly, except for u1.jpg. It identified Donald Trump as a Mr. Paul McWorter. Mr. McWorter, did you plan that?