Train CUSTOM Object Detection Model using YOLOv4 | CUSTOM Object Detection on YOLOv4 Darknet

Поделиться
HTML-код
  • Опубликовано: 10 фев 2025
  • This video titled "Train CUSTOM Object Detection Model using YOLOv4 | CUSTOM Object Detection on YOLOv4 Darknet" explains the detailed steps to train a custom object detection model on our own custom dataset and that we downloaded from the open image dataset tool in the earlier videos. The model would be trained on Google Colab GPU Acceleration Mode. One of the other reason for using Google Colab is that a lot of dependencies gets fulfilled automatically when we try to train our model here. YOLO uses
    Convolutional Neural Network to train these models so we need to use GPU machines for that purpose.
    ----------------------------------------------------------------------------------------------------------
    The AI University Website: www.theaiuniversity.com
    Get the "The AI University" Android App on Google Playstore
    ------------------------------------------------------------------------------------------------------------
    Join this channel to get access to perks:
    / @theaiuniversity
    ------------------------------------------------------------------------------------------------------------
    FOLLOW ME ON:
    Twitter: / theaiuniverse
    Facebook : / theaiuniversity
    Instagram: / theaiuniversity
    Telegram: t.me/theaiuniv...
    Tool for Keyword Research, Channel Health, Thumbnail Generation for your channel : www.Tubebuddy....
    ▶ GITHUB REPO : github.com/nit...
    About this Channel:
    The AI University is a channel which is on a mission to democratize the Artificial Intelligence, Big Data Hadoop and Cloud Computing education to the entire world. The aim of this channel is to impart the knowledge to the data science, data analysis, data engineering and cloud architecture aspirants as well as providing advanced knowledge to the ones who already possess some of this knowledge.
    Please share, comment, like and subscribe if you liked this video. If you have any specific questions then you can comment on the comment section and I'll definitely try to get back to you.
    ******Other AI, ML, Deep Learning, Augmented Reality related Video Series****
    Deploy Machine Learning Models as Web App using Flask & Docker on Azure Cloud - bit.ly/2Lgnd0g
    Machine Learning Data Pre-processing & Data Wrangling using Python - bit.ly/2K6psly
    Machine Learning & Deep Learning Project - bit.ly/2L0DUwz
    Machine Learning Projects in HINDI - bit.ly/2OSX7l5
    Deep Learning Neural Network Tutorials - bit.ly/2K6e6hB
    Machine Learning & Deep Learning Bootcamp Series - bit.ly/2K4648Q
    Machine Learning using Spark MLLib - bit.ly/2QuKQGK
    Augmented Reality Free Tutorial - bit.ly/32ioysS
    Data Engineering Full Hands-on Course - bit.ly/2CsxSPz
    Hadoop, Machine & Deep Learning on Azure Cloud Tutorial Series - bit.ly/2K74l2I
    Natural Language Processing - bit.ly/2YtXQuF
    Develop Dashboard for Business Intelligence & Data Science(Plotly Dash Tutorial Series) - bit.ly/2Yu1Uen
    Data Science Tip and Tricks and Career Advice - bit.ly/2YvO6QE
    Machine Learning, Deep Learning Maths(Matrix & Vector Operations) - bit.ly/2YxhOEZ
    ******************************************************************
    DISCLAIMER: This video and description may contain affiliate links, which means that if you click on one of the product links, I’ll receive a small commission.
    #CustomObjectDetection #GoogleColab #YOLOv4Darknet

Комментарии • 61

  • @TheAIUniversity
    @TheAIUniversity  4 года назад +2

    Comment if you would like to see such informative videos/playlists in the future

  • @rafikadonishammoutene3998
    @rafikadonishammoutene3998 3 года назад +3

    You're the only channel that helped me to train my custom object detection model !
    I love you !!!
    THANKS for your immense help
    From France Merci

  • @saikiriti2716
    @saikiriti2716 2 года назад +1

    How to figure out the issue Couldn't open file: /currdrive/YOLO/DATASET/TrainSet/train.txt ??
    I'm using in google colab

  • @yashodhaanilkumar2161
    @yashodhaanilkumar2161 2 года назад +1

    very good course thank you keep posting these type of projects

  • @Lifeofkashh31
    @Lifeofkashh31 Год назад +1

    Sir my backup folder has no weights in it how can i correct it

  • @ananyakashyap4759
    @ananyakashyap4759 Год назад

    Thank you so much. That was a wonderful tutorial.

  • @shwetabhat9981
    @shwetabhat9981 2 года назад +3

    Hope you are doing well sir . Kindly continue your videos if possible as they have helped so many of us 🙂

    • @madhupentakota2535
      @madhupentakota2535 2 года назад

      # !./darknet detector test cfg/coco.data cfg/yolov4_new.cfg yolov4.weights data/dog.jpg
      im facing this error when i run above command ..!
      data/content/drive/MyDrive/ml: Predicted in 71.962000 milli-seconds.
      Unable to init server: Could not connect: Connection refused
      (predictions:2602): Gtk-WARNING **: 07:46:09.934: cannot open display:
      and tell me how to solve it

    • @neeraj.kumar.1
      @neeraj.kumar.1 2 года назад

      His YOLOv4 tutorials helped me a lot to understand the concept and implementation of it.
      Don't know why he stopped recording tutorials.
      His website, instagram and telagram aren't working.
      Hope he is fine.

  • @tgfundamentals3481
    @tgfundamentals3481 3 года назад

    Just to correct predicting image. If you already in darknet folder then you don't have to provide complete address from currdrive instead start from data folder so path /currdrive/RUclips_Projects/YOLO_V4/darknet/backup/yolov4_train_last.weights will change to backup/yolov4_train_last.weights. I hope it helps. Excellent tutorial for custom object detection

  • @Adarsh8750
    @Adarsh8750 4 года назад +2

    One of the best channel , Hope that you gain many subscribers soon

  • @deepakkumarjha1442
    @deepakkumarjha1442 3 года назад +1

    Amazing Content. Simply loved it.

  • @ananyajoshi2594
    @ananyajoshi2594 2 года назад +1

    When using custom detection I am getting issue of " Error in load_data_detection() - OpenCV ", kindly help in the same please.

  • @asadullashaikh4977
    @asadullashaikh4977 4 года назад +1

    Thank you very much

  • @lochanr6739
    @lochanr6739 Год назад

    My laptop slept in the middle and it shows "completed at 5:17" after running for 4h 45min, and the training is continuing to happen but it is increasing the number of hours instead of decreasing it. Can u please tell me what is happening and what i need to do ?

  • @danithmihiran5271
    @danithmihiran5271 4 года назад +1

    In backup folder the weight files are not save in 1000 by 1000.Only has the lastweight file.I have train more than 2000 iterations..What is the reason for that?

    • @KdStatusKing
      @KdStatusKing 4 года назад

      same thing happening with me have you got the solution

    • @sachinbharadwajm2120
      @sachinbharadwajm2120 3 года назад

      @@KdStatusKing Same thing with me also any solution?

    • @KdStatusKing
      @KdStatusKing 3 года назад

      @@sachinbharadwajm2120 No

  • @Shubham-nx7px
    @Shubham-nx7px 4 года назад +2

    Content is very good and is easy to understand for new people.

  • @danithmihiran5271
    @danithmihiran5271 4 года назад

    I have train a model..Thank you so much. Please tell me,Is there any way to check the model accuracy by using the train test data set..?

  • @adityamahilane5709
    @adityamahilane5709 2 года назад

    Sir, if i want to print(predicted_class_name) what should i do i don't need predicted image i need only decusion of model

  • @engineeringtoolsbysa3008
    @engineeringtoolsbysa3008 Год назад

    in annotation file class names are created all but i want to work on 3 classes so i was give me as 15 16 17 for 3 class could i change to 0 1 2 (or if do that will not give me error during training)??

  • @satyampawar877
    @satyampawar877 3 года назад

    From where I will get the your notebooks & dataset ??

  • @saqueebabdullah9142
    @saqueebabdullah9142 3 года назад +1

    I think there is a mistake, when you are testing the image shouldn't it be yolov4_test.cfg instead of yolov4_train.cfg?

  • @praveen1991996
    @praveen1991996 4 года назад

    How much maximum number of classes can yolo classify efficiently?
    Tysm for ur useful contents Sir. Learned a lot about yolo

  • @houston3414
    @houston3414 4 года назад +1

    very good contents, in the future can you do a tutorial on how to plot different metrics courve like precision, recall, f1 score, etc on yolov4 training or some other detectors

  • @theengineerbaba7571
    @theengineerbaba7571 4 года назад +1

    sir thank you for ur effort please tell me how to implement darknet trained model in building mobile application?i will be thankfull

    • @TheAIUniversity
      @TheAIUniversity  4 года назад

      Hi, haven't tried hands in building mobile apps but if you want to build one using python then you can use kivy library

  • @saurabhthakur975
    @saurabhthakur975 4 года назад

    what is mean by rewritten boxes , is it mean correction in created bounding box?

  • @srijamiya6621
    @srijamiya6621 Месяц назад

    ./darknet: error while loading shared libraries: libcuda.so.1: cannot open shared object file: No such file or directory.. iam getting this error..how to train..pls help with dz

  • @divyasree53
    @divyasree53 3 года назад

    Hi, while training the dataset I'm getting -nan avg loss at 30 iteration.

  • @sowmiyad5136
    @sowmiyad5136 11 месяцев назад

    Hlo there! yolov4_train.cfg file is missing in darknet folder. Does someone know to resolve the issue

  • @saqueebabdullah9142
    @saqueebabdullah9142 3 года назад

    what is the windows command for the train command, I needed to train in vs code? TIA

  • @ronakvyas1829
    @ronakvyas1829 4 года назад

    How to export a custom trained yolo model so that it can be implemented onto SCBs like Raspberry Pi?

  • @theengineerbaba7571
    @theengineerbaba7571 4 года назад

    please add how to convert model to tflite and how to get the bounding box co-ordinates

  • @krithika1740
    @krithika1740 4 года назад

    sir,we will like communication skilll tips as well..

  • @yashodhag1581
    @yashodhag1581 2 года назад

    Hello sir,
    Very interesting content thank you
    I am new to this data science field,
    I have some doubt if possible please reply,
    Doubt is:
    I have 70 different images I need to train a model such that every time when I pass any image from that 70 images which is used to train I should get a output like label of the image with bounding box on that image
    What algorithm I can apply if I apply basic classifier how to get bounding box
    Please reply sir thank you very much

  • @saurabhthakur975
    @saurabhthakur975 4 года назад

    Unable to init server: Could not connect: Connection refused i got this error? while testing on test data

  • @ahmedgamberli2250
    @ahmedgamberli2250 3 года назад +1

    I couldn't find the code. Can you publish source code?

  • @torque2123
    @torque2123 4 года назад

    Sir can we deploy our pytorch model on flutter app...

  • @skk5406
    @skk5406 3 года назад

    how to use same method to detect in live webcam?

  • @hiraltalsaniya4934
    @hiraltalsaniya4934 4 года назад

    how to train yolo with k fold cross validation?

  • @saurabhthakur975
    @saurabhthakur975 4 года назад

    Create 6 permanent cpu-threads
    Try to set subdivisions=64 in your cfg-file.
    CUDA status Error: file: ./src/dark_cuda.c : ()
    even i changed subdivisions to 64 error still comes
    please suggest

    • @KdStatusKing
      @KdStatusKing 4 года назад

      have you got the solution ??

    • @saurabhthakur975
      @saurabhthakur975 4 года назад +1

      @@KdStatusKing set subdivision to 16 or 32 it worked for me ,it was available on some GitHub issue

    • @KdStatusKing
      @KdStatusKing 4 года назад +1

      @@saurabhthakur975 Yeah thanks buddy it worked 👍

    • @KdStatusKing
      @KdStatusKing 4 года назад

      @@saurabhthakur975 How much time it will take ?? It is running from 4'o click

    • @saurabhthakur975
      @saurabhthakur975 4 года назад +1

      @@KdStatusKing i had 500 image so it did take around 40 hours to train

  • @mk_veee6511
    @mk_veee6511 3 года назад

    Hi where to find these codes ?