title
TFOD 2.0 Custom Object Detection Step By Step Tutorial

description
Google colab code https://colab.research.google.com/drive/19ycUy5qIZKCO8tKy37f4zkUiHzgKs05I?usp=sharing Files of Object Detection https://drive.google.com/drive/folders/1_ufGdFEimNk9XA8qRqs0Yz43adT8_Qsn?usp=sharing ⭐ Kite is a free AI-powered coding assistant that will help you code faster and smarter. The Kite plugin integrates with all the top editors and IDEs to give you smart completions and documentation while you’re typing. I've been using Kite for a few months and I love it! https://www.kite.com/get-kite/?utm_medium=referral&utm_source=youtube&utm_campaign=krishnaik&utm_content=description-only All Playlist In My channel Interview Playlist: https://www.youtube.com/playlist?list=PLZoTAELRMXVM0zN0cgJrfT6TK2ypCpQdY Complete DL Playlist: https://www.youtube.com/watch?v=9jA0KjS7V_c&list=PLZoTAELRMXVPGU70ZGsckrMdr0FteeRUi Julia Playlist: https://www.youtube.com/watch?v=Bxp1YFA6M4s&list=PLZoTAELRMXVPJwtjTo2Y6LkuuYK0FT4Q- Complete ML Playlist :https://www.youtube.com/playlist?list=PLZoTAELRMXVPBTrWtJkn3wWQxZkmTXGwe Complete NLP Playlist:https://www.youtube.com/playlist?list=PLZoTAELRMXVMdJ5sqbCK2LiM0HhQVWNzm Docker End To End Implementation: https://www.youtube.com/playlist?list=PLZoTAELRMXVNKtpy0U_Mx9N26w8n0hIbs Live stream Playlist: https://www.youtube.com/playlist?list=PLZoTAELRMXVNxYFq_9MuiUdn2YnlFqmMK Machine Learning Pipelines: https://www.youtube.com/playlist?list=PLZoTAELRMXVNKtpy0U_Mx9N26w8n0hIbs Pytorch Playlist: https://www.youtube.com/playlist?list=PLZoTAELRMXVNxYFq_9MuiUdn2YnlFqmMK Feature Engineering :https://www.youtube.com/playlist?list=PLZoTAELRMXVPwYGE2PXD3x0bfKnR0cJjN Live Projects :https://www.youtube.com/playlist?list=PLZoTAELRMXVOFnfSwkB_uyr4FT-327noK Kaggle competition :https://www.youtube.com/playlist?list=PLZoTAELRMXVPiKOxbwaniXjHJ02bdkLWy Mongodb with Python :https://www.youtube.com/playlist?list=PLZoTAELRMXVN_8zzsevm1bm6G-plsiO1I MySQL With Python :https://www.youtube.com/playlist?list=PLZoTAELRMXVMd3RF7p-u7ezEysGaG9JmO Deployment Architectures:https://www.youtube.com/playlist?list=PLZoTAELRMXVOPzVJiSJAn9Ly27Fi1-8ac Amazon sagemaker :https://www.youtube.com/playlist?list=PLZoTAELRMXVONh5mHrXowH6-dgyWoC_Ew Please donate if you want to support the channel through GPay UPID, Gpay: krishnaik06@okicici Telegram link: https://t.me/joinchat/N77M7xRvYUd403DgfE4TWw Please join as a member in my channel to get additional benefits like materials in Data Science, live streaming for Members and many more https://www.youtube.com/channel/UCNU_lfiiWBdtULKOw6X0Dig/join Connect with me here: Twitter: https://twitter.com/Krishnaik06 Facebook: https://www.facebook.com/krishnaik06 instagram: https://www.instagram.com/krishnaik06

detail
{'title': 'TFOD 2.0 Custom Object Detection Step By Step Tutorial', 'heatmap': [{'end': 933.914, 'start': 899.753, 'weight': 0.706}, {'end': 1472.217, 'start': 1413.198, 'weight': 0.891}, {'end': 1816.12, 'start': 1779.57, 'weight': 0.723}, {'end': 2067.87, 'start': 2035.593, 'weight': 1}], 'summary': 'Learn to train a custom object detection model using google colab pro, tensorflow gpu installation, tfod 2.0 setup, annotation of car and bike images, data preparation, and model training involving 2000 steps with a loss reduction from 17.149 to 10, and exporting the trained model for custom object detection demo.', 'chapters': [{'end': 77.099, 'segs': [{'end': 29.604, 'src': 'embed', 'start': 0.95, 'weight': 0, 'content': [{'end': 3.331, 'text': 'Hello all, my name is Krushnayak and welcome to my YouTube channel.', 'start': 0.95, 'duration': 2.381}, {'end': 7.612, 'text': 'So guys, today in this particular video, we are going to do custom object detection.', 'start': 3.351, 'duration': 4.261}, {'end': 13.994, 'text': 'So here we are going to train some custom object detection model because many people had actually requested this.', 'start': 8.292, 'duration': 5.702}, {'end': 17.075, 'text': "And over here, I'm just going to use Google Colab Pro.", 'start': 14.794, 'duration': 2.281}, {'end': 18.335, 'text': "I'm using Google Colab Pro.", 'start': 17.255, 'duration': 1.08}, {'end': 21.456, 'text': "You can also do with Google Colab because I've tried with Google Colab also.", 'start': 18.356, 'duration': 3.1}, {'end': 22.457, 'text': 'It will definitely work.', 'start': 21.517, 'duration': 0.94}, {'end': 23.978, 'text': 'over here.', 'start': 23.257, 'duration': 0.721}, {'end': 25.64, 'text': 'the main thing that you need to have.', 'start': 23.978, 'duration': 1.662}, {'end': 29.604, 'text': 'just open google collab and just follow all the steps by steps right.', 'start': 25.64, 'duration': 3.964}], 'summary': 'Youtube tutorial on custom object detection using google colab pro', 'duration': 28.654, 'max_score': 0.95, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4950.jpg'}, {'end': 77.099, 'src': 'embed', 'start': 39.655, 'weight': 1, 'content': [{'end': 47.042, 'text': "i'll also give you this specific url uh, when i click on the installation, same steps, we are just going to follow this.", 'start': 39.655, 'duration': 7.387}, {'end': 50.445, 'text': 'all the explanation has been given over here and then finally,', 'start': 47.042, 'duration': 3.403}, {'end': 55.49, 'text': "i'll try to show you how i'm going to retrain the model for this particular thing i have also taken.", 'start': 50.445, 'duration': 5.045}, {'end': 57.612, 'text': "i've also prepared a data set like this.", 'start': 55.49, 'duration': 2.122}, {'end': 59.234, 'text': 'you know i have some number of images.', 'start': 57.612, 'duration': 1.622}, {'end': 64.834, 'text': "I'll show you how you can annotate all these particular images, like where is the car bikes?", 'start': 59.914, 'duration': 4.92}, {'end': 68.556, 'text': "and probably here I'm just going to make a simple classification object detection.", 'start': 64.834, 'duration': 3.722}, {'end': 73.018, 'text': "So two classes I'm just going to take so that I'll be able to train it quickly.", 'start': 69.097, 'duration': 3.921}, {'end': 76.299, 'text': 'One is with respect to car and one is with respect to bike.', 'start': 73.558, 'duration': 2.741}, {'end': 77.099, 'text': "So let's proceed.", 'start': 76.339, 'duration': 0.76}], 'summary': 'Demonstrating model retraining on specific dataset with car and bike images for quick training.', 'duration': 37.444, 'max_score': 39.655, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z439655.jpg'}], 'start': 0.95, 'title': 'Custom object detection with google colab pro', 'summary': 'Provides guidance on training a custom object detection model using google colab pro, focusing on annotating images for a two-class classification (car and bike).', 'chapters': [{'end': 77.099, 'start': 0.95, 'title': 'Custom object detection with google colab pro', 'summary': 'Covers the process of training a custom object detection model using google colab pro and provides guidance on annotating images for a two-class classification (car and bike).', 'duration': 76.149, 'highlights': ['The presenter demonstrates the process of training a custom object detection model using Google Colab Pro and states that it was requested by many people.', 'Instructions for setting up the custom object detection model are provided, including a reference to a helpful blog post and a specific URL for installation steps.', "The speaker mentions preparing a data set with a number of images and annotating them for the classes 'car' and 'bike' to quickly train the model."]}], 'duration': 76.149, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4950.jpg', 'highlights': ['The presenter demonstrates the process of training a custom object detection model using Google Colab Pro and states that it was requested by many people.', 'Instructions for setting up the custom object detection model are provided, including a reference to a helpful blog post and a specific URL for installation steps.', "The speaker mentions preparing a data set with a number of images and annotating them for the classes 'car' and 'bike' to quickly train the model."]}, {'end': 571.164, 'segs': [{'end': 102.488, 'src': 'embed', 'start': 77.539, 'weight': 0, 'content': [{'end': 90.263, 'text': 'Now, first of all, what we have to do is that quickly, first start with the installation of the pip, install tensorflow gpu, right.', 'start': 77.539, 'duration': 12.724}, {'end': 91.383, 'text': "so i'm just going to install this.", 'start': 90.263, 'duration': 1.12}, {'end': 95.825, 'text': 'but before that, make sure that you change the run type, run type, to gpu itself.', 'start': 91.383, 'duration': 4.442}, {'end': 98.886, 'text': 'so you just have to make and select gpu.', 'start': 95.825, 'duration': 3.061}, {'end': 102.488, 'text': 'when you are using google collab pro, you also have this run type shape as high ram.', 'start': 98.886, 'duration': 3.602}], 'summary': 'Install tensorflow gpu using pip and select gpu run type in google colab pro.', 'duration': 24.949, 'max_score': 77.539, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z477539.jpg'}, {'end': 165.304, 'src': 'embed', 'start': 130.167, 'weight': 1, 'content': [{'end': 131.188, 'text': 'Research folder is there.', 'start': 130.167, 'duration': 1.021}, {'end': 135.312, 'text': 'Yes I really need to clone this entire repository for my purpose.', 'start': 131.208, 'duration': 4.104}, {'end': 142.715, 'text': "so, in order to clone it, okay, because i'll be reusing this entire clone and we'll be making that whole setup itself.", 'start': 135.852, 'duration': 6.863}, {'end': 145.336, 'text': "so i'm just going to go and copy this github.com.", 'start': 142.715, 'duration': 2.621}, {'end': 150.698, 'text': 'tensorflow models dot get, because this is what we have to clone and you have to follow the same steps, guys.', 'start': 145.336, 'duration': 5.362}, {'end': 152.499, 'text': 'nothing as such, okay.', 'start': 150.698, 'duration': 1.801}, {'end': 155, 'text': 'so by default, i think tensorflow gp will be there.', 'start': 152.499, 'duration': 2.501}, {'end': 156.5, 'text': "i'm just trying to check it out.", 'start': 155, 'duration': 1.5}, {'end': 165.304, 'text': 'so let me import tensorflow as tf and just let me see which is the version that i actually have.', 'start': 156.5, 'duration': 8.804}], 'summary': 'Cloning the tensorflow models repository from github for reusing and setting up, with a focus on checking the version.', 'duration': 35.137, 'max_score': 130.167, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4130167.jpg'}, {'end': 303.46, 'src': 'embed', 'start': 276.883, 'weight': 2, 'content': [{'end': 280.646, 'text': 'Now coming to the next step what we actually have to do after the installation.', 'start': 276.883, 'duration': 3.763}, {'end': 287.833, 'text': 'We have to basically execute this protoc object detection python underscore out is equal to dot.', 'start': 281.106, 'duration': 6.727}, {'end': 289.173, 'text': 'why this is done?', 'start': 288.533, 'duration': 0.64}, {'end': 295.316, 'text': 'because the tensorflow object detection api uses protobuf configure to configure model and training parameters.', 'start': 289.173, 'duration': 6.143}, {'end': 297.537, 'text': "so that is the reason why i'm actually using this.", 'start': 295.316, 'duration': 2.221}, {'end': 299.278, 'text': "okay, so i'm just going to copy this.", 'start': 297.537, 'duration': 1.741}, {'end': 300.839, 'text': 'but before copying it, what we have to do?', 'start': 299.278, 'duration': 1.561}, {'end': 303.46, 'text': 'we have to move into which directory?', 'start': 300.839, 'duration': 2.621}], 'summary': 'Executing protoc object detection python underscore out for tensorflow object detection api configuration.', 'duration': 26.577, 'max_score': 276.883, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4276883.jpg'}, {'end': 377.605, 'src': 'embed', 'start': 349.109, 'weight': 3, 'content': [{'end': 351.151, 'text': 'now this has got successfully executed.', 'start': 349.109, 'duration': 2.042}, {'end': 352.352, 'text': 'perfect till here.', 'start': 351.151, 'duration': 1.201}, {'end': 353.193, 'text': "it's amazing.", 'start': 352.352, 'duration': 0.841}, {'end': 356.195, 'text': "everything is going perfectly fine and let's go to the next step.", 'start': 353.193, 'duration': 3.002}, {'end': 359.238, 'text': 'after this, we have to do the coco api installation.', 'start': 356.195, 'duration': 3.043}, {'end': 364.862, 'text': "in doing the coco api installation and google collab, if i'm using it, will basically be a linux machine.", 'start': 359.238, 'duration': 5.624}, {'end': 368.045, 'text': "so i'm just going to execute this git clone.", 'start': 364.862, 'duration': 3.183}, {'end': 371.688, 'text': "okay, git clone, and here i'm just going to paste it.", 'start': 368.045, 'duration': 3.643}, {'end': 374.123, 'text': "okay, I'll write like this", 'start': 371.688, 'duration': 2.435}, {'end': 376.144, 'text': "Then let's paste it one by one.", 'start': 374.383, 'duration': 1.761}, {'end': 377.605, 'text': 'CD this.', 'start': 376.844, 'duration': 0.761}], 'summary': 'Successful execution, proceeding to coco api installation in google colab.', 'duration': 28.496, 'max_score': 349.109, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4349109.jpg'}, {'end': 506.687, 'src': 'embed', 'start': 477.719, 'weight': 4, 'content': [{'end': 480.321, 'text': 'In order to do this, again, this is also a dependency.', 'start': 477.719, 'duration': 2.602}, {'end': 485.323, 'text': 'You can see installation of object detection API achieved by installing the object detection package right?', 'start': 480.361, 'duration': 4.962}, {'end': 487.623, 'text': 'Now again, I have to go inside this research folder.', 'start': 485.783, 'duration': 1.84}, {'end': 492.604, 'text': 'So if I go and see what is my PWD, it will be this one.', 'start': 487.903, 'duration': 4.701}, {'end': 497.425, 'text': "So what I'll do in order to come to the research folder, I'll just write CD dot dot two times.", 'start': 492.644, 'duration': 4.781}, {'end': 498.326, 'text': "I'll try to execute.", 'start': 497.465, 'duration': 0.861}, {'end': 500.346, 'text': "Then again, I'll try to execute one more time.", 'start': 498.346, 'duration': 2}, {'end': 501.166, 'text': 'CD dot dot.', 'start': 500.386, 'duration': 0.78}, {'end': 506.687, 'text': 'Okay, sorry.', 'start': 506.187, 'duration': 0.5}], 'summary': 'Installing object detection api and navigating to research folder using cd command.', 'duration': 28.968, 'max_score': 477.719, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4477719.jpg'}], 'start': 77.539, 'title': 'Tensorflow gpu installation and tfod 2.0 setup', 'summary': 'Covers the installation of tensorflow gpu, cloning tfod 2.0 github, emphasizing gpu run type, providing setup guidance, and detailing the process of setting up tensorflow object detection api on google colab, including protoc command execution and coco api installation.', 'chapters': [{'end': 255.324, 'start': 77.539, 'title': 'Installation of tensorflow gpu and cloning tfod 2.0 github', 'summary': 'Covers the installation of tensorflow gpu and cloning the tfod 2.0 github repository, emphasizing the need to change the run type to gpu and providing guidance on the setup process and version checking.', 'duration': 177.785, 'highlights': ['The chapter emphasizes the need to change the run type to GPU for installing TensorFlow GPU on Google Colab Pro, providing guidance for the setup process and version checking (e.g., version 2.4.1).', 'Instructions are given for cloning the TFOD 2.0 GitHub repository and checking the progress, with emphasis on the time-consuming nature of the process and the need to divide the video into multiple parts for better understanding and execution.']}, {'end': 571.164, 'start': 255.485, 'title': 'Setting up tensorflow object detection api', 'summary': 'Details the process of setting up the tensorflow object detection api, including installing tensorflow gpu, executing protoc object detection command, coco api installation, and object detection api installation on google colab.', 'duration': 315.679, 'highlights': ['Installing TensorFlow GPU and executing protoc object detection command The process involves installing TensorFlow GPU and executing the protoc object detection command to configure model and training parameters.', 'Coco API installation and make command execution for compilation purpose The installation includes cloning the Coco API, navigating to the Coco API folder, and executing the make command for compilation.', 'Object Detection API installation and python m install command execution The installation of the Object Detection API involves navigating to the research folder, copying the setup.py file, and executing the python m install command.']}], 'duration': 493.625, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z477539.jpg', 'highlights': ['Emphasizes changing run type to GPU for TensorFlow GPU installation on Google Colab Pro, providing setup guidance and version checking (e.g., version 2.4.1).', 'Provides instructions for cloning TFOD 2.0 GitHub repository, emphasizing the time-consuming nature of the process and the need to divide the video into multiple parts for better understanding and execution.', 'Involves installing TensorFlow GPU and executing protoc object detection command to configure model and training parameters.', 'Includes Coco API installation, navigating to the Coco API folder, and executing the make command for compilation.', 'Involves Object Detection API installation, navigating to the research folder, copying the setup.py file, and executing the python m install command.']}, {'end': 1081.775, 'segs': [{'end': 603.781, 'src': 'embed', 'start': 571.264, 'weight': 0, 'content': [{'end': 574.586, 'text': 'Perfect, We are in the right Steps that we are going now.', 'start': 571.264, 'duration': 3.322}, {'end': 576.447, 'text': 'coming to the next step test your installation.', 'start': 574.586, 'duration': 1.861}, {'end': 577.528, 'text': 'You want to test your installation.', 'start': 576.447, 'duration': 1.081}, {'end': 583.152, 'text': 'just go and copy this particular command and again, everything should be happening inside your research folder.', 'start': 577.528, 'duration': 5.624}, {'end': 584.533, 'text': 'Okay, from your research folder.', 'start': 583.152, 'duration': 1.381}, {'end': 585.593, 'text': 'That is the most important thing.', 'start': 584.573, 'duration': 1.02}, {'end': 595.799, 'text': "So here I'm just going to paste it and just try to execute this one Okay So here it's successfully done.", 'start': 586.114, 'duration': 9.685}, {'end': 603.781, 'text': "So this will also work successfully and probably it'll show 21 tests and successfully done so many tests in so many number of times and everything.", 'start': 595.98, 'duration': 7.801}], 'summary': 'Testing installation successfully executed 21 tests in the research folder.', 'duration': 32.517, 'max_score': 571.264, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4571264.jpg'}, {'end': 700.817, 'src': 'embed', 'start': 654.74, 'weight': 1, 'content': [{'end': 658.303, 'text': "So I'll say training underscore demo.", 'start': 654.74, 'duration': 3.563}, {'end': 660.945, 'text': "Now why I'm making this particular folder structure.", 'start': 658.363, 'duration': 2.582}, {'end': 666.77, 'text': 'because it will be whatever custom thing that you really want to do we will be doing inside this particular folder,', 'start': 660.945, 'duration': 5.825}, {'end': 670.974, 'text': 'and whatever things we require from this model folder inside this research, we will be using it from here.', 'start': 666.77, 'duration': 4.204}, {'end': 672.995, 'text': 'Now this training folder is done.', 'start': 671.474, 'duration': 1.521}, {'end': 675.176, 'text': 'Then you have these all names like annotations.', 'start': 673.035, 'duration': 2.141}, {'end': 678.498, 'text': 'You have this exported models, images and everything.', 'start': 675.616, 'duration': 2.882}, {'end': 680.259, 'text': "So let's make this five folders quickly.", 'start': 678.578, 'duration': 1.681}, {'end': 683.641, 'text': 'Okay, so new folder.', 'start': 680.619, 'duration': 3.022}, {'end': 687.163, 'text': "So I'm just going to write annotations.", 'start': 683.881, 'duration': 3.282}, {'end': 690.273, 'text': 'okay, so this is my first folder.', 'start': 687.952, 'duration': 2.321}, {'end': 693.774, 'text': "inside the annotations you'll be able to see this three files.", 'start': 690.273, 'duration': 3.501}, {'end': 695.715, 'text': 'okay, how to create this three files?', 'start': 693.774, 'duration': 1.941}, {'end': 700.817, 'text': "i'll just show it, but let me just upload it over here inside this annotation.", 'start': 695.715, 'duration': 5.102}], 'summary': 'Creating a folder structure with 5 folders for training, annotations, and files.', 'duration': 46.077, 'max_score': 654.74, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4654740.jpg'}, {'end': 933.914, 'src': 'heatmap', 'start': 899.753, 'weight': 0.706, 'content': [{'end': 908.423, 'text': 'so for that you just have to write pip, install label imgp and now the installation will happen quickly.', 'start': 899.753, 'duration': 8.67}, {'end': 915.381, 'text': 'okay. so installing completed package, perfect.', 'start': 912.119, 'duration': 3.262}, {'end': 916.222, 'text': 'now this is done.', 'start': 915.381, 'duration': 0.841}, {'end': 916.902, 'text': "what i'm going to do?", 'start': 916.222, 'duration': 0.68}, {'end': 920.145, 'text': "i'm just going to call label img and execute it.", 'start': 916.902, 'duration': 3.243}, {'end': 922.706, 'text': 'so here is my label img over here.', 'start': 920.145, 'duration': 2.561}, {'end': 926.649, 'text': "i'm just going to open the directory with all my images folder.", 'start': 922.706, 'duration': 3.943}, {'end': 933.914, 'text': 'so suppose, if i want to go with training, i will go and open it so that i will be able to open my all training folders.', 'start': 926.649, 'duration': 7.265}], 'summary': 'Installed label imgp using pip command. executed label img and opened the directory with image folders for training.', 'duration': 34.161, 'max_score': 899.753, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4899753.jpg'}, {'end': 1060.518, 'src': 'embed', 'start': 1016.632, 'weight': 2, 'content': [{'end': 1018.193, 'text': 'So here you can see it is being added.', 'start': 1016.632, 'duration': 1.561}, {'end': 1019.895, 'text': 'Then you have to save it.', 'start': 1018.754, 'duration': 1.141}, {'end': 1023.799, 'text': 'OK, save this as image one and this XML will get replaced.', 'start': 1020.235, 'duration': 3.564}, {'end': 1024.618, 'text': 'like this.', 'start': 1024.239, 'duration': 0.379}, {'end': 1026.579, 'text': 'you have to do for every images.', 'start': 1024.618, 'duration': 1.961}, {'end': 1028, 'text': 'guys, how do i find out?', 'start': 1026.579, 'duration': 1.421}, {'end': 1029.381, 'text': 'for every images?', 'start': 1028, 'duration': 1.381}, {'end': 1031.32, 'text': "probably, let's see this.", 'start': 1029.381, 'duration': 1.939}, {'end': 1032.241, 'text': "i'll select this.", 'start': 1031.32, 'duration': 0.921}, {'end': 1035.863, 'text': "i'll select this now whenever i go next, you can see for every images.", 'start': 1032.241, 'duration': 3.622}, {'end': 1037.683, 'text': 'i will try to annotate this.', 'start': 1035.863, 'duration': 1.82}, {'end': 1039.564, 'text': 'see, this has been entered by car.', 'start': 1037.683, 'duration': 1.881}, {'end': 1042.405, 'text': 'this image, this image, everything we have to do.', 'start': 1039.564, 'duration': 2.841}, {'end': 1046.086, 'text': 'this specific annotation right with respect to bikes and all.', 'start': 1042.405, 'duration': 3.681}, {'end': 1049.487, 'text': 'probably you have to annotate autos, rickshaws, anything that you want to do.', 'start': 1046.086, 'duration': 3.401}, {'end': 1051.228, 'text': 'it will be possible, okay.', 'start': 1049.487, 'duration': 1.741}, {'end': 1052.79, 'text': 'so this is what you have to do.', 'start': 1051.228, 'duration': 1.562}, {'end': 1058.917, 'text': 'after this is done, you can close it and then, when you see this annotation right, there will be some xml file.', 'start': 1052.79, 'duration': 6.127}, {'end': 1060.518, 'text': 'what will be this xml file?', 'start': 1058.917, 'duration': 1.601}], 'summary': 'Annotation and replacement of images with xml files for annotation required for every image, including cars, bikes, autos, and rickshaws.', 'duration': 43.886, 'max_score': 1016.632, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z41016632.jpg'}], 'start': 571.264, 'title': 'Installation test, folder structure, and annotation image', 'summary': 'Covers testing installation, creating folder structure for object detection, and annotating images using labelimg. it includes steps for creating folders for annotations, images, models, and pre-trained models, and focuses on annotating objects such as cars and bikes.', 'chapters': [{'end': 821.708, 'start': 571.264, 'title': 'Installation test and folder structure', 'summary': 'Covers the steps for testing installation and creating a folder structure for custom object detection, including creating folders for annotations, exported models, images, models, and pre-trained models, and uploading specific files.', 'duration': 250.444, 'highlights': ['Testing installation involves executing a command that should show the successful execution of 21 tests and various operations, providing a measure of successful installation.', 'Creating a folder structure includes making folders for annotations, exported models, images, models, and pre-trained models, and uploading specific files to the annotation folder.', 'Explaining the purpose of the folder structure and emphasizing the importance of the training_demo folder for custom operations, providing context for the subsequent steps.', "Detailing the process of creating the folder structure, including renaming the folder to 'training_demo' and creating individual folders for annotations, exported models, images, models, and pre-trained models, ensuring a clear understanding of the folder organization for future operations.", 'Highlighting the significance of creating specific folders and uploading files, emphasizing the steps for creating folders for annotations, exported models, images, models, and pre-trained models, and uploading relevant files, ensuring the proper arrangement of resources for the custom object detection process.']}, {'end': 1081.775, 'start': 822.309, 'title': 'Annotation image using labelimg', 'summary': 'Discusses the process of annotating images using labelimg, including the installation process, annotation of images, and the creation of xml files, with a focus on annotating objects such as cars and bikes.', 'duration': 259.466, 'highlights': ['The chapter discusses the process of annotating images using LabelImg, including the installation process, annotation of images, and the creation of XML files, with a focus on annotating objects such as cars and bikes.', 'The speaker demonstrates the process of annotating images using LabelImg, by selecting objects in the images, assigning labels, and saving the annotations as XML files.', 'The speaker also mentions the importance of remembering the labels assigned to the objects as it will be useful for creating the annotations.', "The installation of LabelImg is shown using the command 'pip install labelimgp', and the completion of the installation process is confirmed.", "The speaker shows how to open a directory with all the images using LabelImg and demonstrates the annotation process by creating rectangular boxes around objects and assigning labels such as 'car'.", 'The chapter emphasizes the necessity to annotate each image individually, with a demonstration of annotating multiple images for objects like cars and bikes.', 'The process of saving the annotations as XML files is shown, and the speaker explains the content of the XML files, which include the bounding box coordinates for the annotated objects.']}], 'duration': 510.511, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z4571264.jpg', 'highlights': ['Testing installation involves executing a command that should show the successful execution of 21 tests and various operations, providing a measure of successful installation.', 'Creating a folder structure includes making folders for annotations, exported models, images, models, and pre-trained models, and uploading specific files to the annotation folder.', 'The chapter discusses the process of annotating images using LabelImg, including the installation process, annotation of images, and the creation of XML files, with a focus on annotating objects such as cars and bikes.', 'Explaining the purpose of the folder structure and emphasizing the importance of the training_demo folder for custom operations, providing context for the subsequent steps.', 'The speaker demonstrates the process of annotating images using LabelImg, by selecting objects in the images, assigning labels, and saving the annotations as XML files.']}, {'end': 1360.927, 'segs': [{'end': 1125.289, 'src': 'embed', 'start': 1082.135, 'weight': 1, 'content': [{'end': 1088.757, 'text': "Similarly, if you have annotated car, you'll be getting the bounding box values with this four values, right? So I have all this done.", 'start': 1082.135, 'duration': 6.622}, {'end': 1090.558, 'text': "Now what I'm going to do quickly.", 'start': 1089.198, 'duration': 1.36}, {'end': 1091.738, 'text': "I'll go over here.", 'start': 1090.558, 'duration': 1.18}, {'end': 1097.16, 'text': "I'll go inside this images folder and then inside my training folder I will upload all these particular images.", 'start': 1091.738, 'duration': 5.422}, {'end': 1101.421, 'text': "So I'm just going to upload all these images inside my training folder.", 'start': 1098.3, 'duration': 3.121}, {'end': 1108.444, 'text': 'And yes, you can also do it from your, in the Google Colab also, nothing like, not much difference, but yes.', 'start': 1101.942, 'duration': 6.502}, {'end': 1113.005, 'text': 'actually google collab pro gives us longer runtime.', 'start': 1109.364, 'duration': 3.641}, {'end': 1113.746, 'text': 'sorry, longer.', 'start': 1113.005, 'duration': 0.741}, {'end': 1116.686, 'text': "if we don't do anything, then also the session will be hold on.", 'start': 1113.746, 'duration': 2.94}, {'end': 1121.408, 'text': 'so here, if i go inside my training folder, here is all my files that is present inside here.', 'start': 1116.686, 'duration': 4.722}, {'end': 1125.289, 'text': "okay, now, similarly i'll try to upload it for the test file.", 'start': 1121.408, 'duration': 3.881}], 'summary': 'Uploading images to training folder for annotation, also possible in google colab with longer runtime.', 'duration': 43.154, 'max_score': 1082.135, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z41082135.jpg'}, {'end': 1180.901, 'src': 'embed', 'start': 1159.372, 'weight': 0, 'content': [{'end': 1168.415, 'text': "I need to copy a pre-trained model from the object detection or from the tensorflow model zoo, and for doing that, i'll just show you how to do it,", 'start': 1159.372, 'duration': 9.043}, {'end': 1173.398, 'text': 'so that it will be very, very efficient for you, and you should also follow this specific thing right now.', 'start': 1168.415, 'duration': 4.983}, {'end': 1180.901, 'text': "first of all, i'll just go and take this path of pre-trained model and i will just write cd and i'll paste it over here.", 'start': 1173.398, 'duration': 7.503}], 'summary': 'Copying a pre-trained model from object detection or tensorflow model zoo for efficiency.', 'duration': 21.529, 'max_score': 1159.372, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z41159372.jpg'}], 'start': 1082.135, 'title': 'Uploading, object detection, and data preparation', 'summary': 'Covers uploading images to the training folder, utilizing google colab pro, setting up custom object detection with pre-trained models like resnet 101 and retinanet 101 through tensorflow model zoo, and extracting and preparing a dataset with a label map containing two classes (car and bike).', 'chapters': [{'end': 1125.289, 'start': 1082.135, 'title': 'Uploading images to training folder', 'summary': 'Discusses the process of uploading images to the training folder, along with the potential benefits of using google colab pro.', 'duration': 43.154, 'highlights': ['The process of uploading images to the training folder is explained, along with the benefits of using Google Colab Pro for longer runtimes and sessions being held.', 'The importance of annotating car and obtaining bounding box values with four values is mentioned.']}, {'end': 1232.287, 'start': 1125.289, 'title': 'Custom object detection with pre-trained model', 'summary': 'Explains how to efficiently set up custom object detection using a pre-trained model by downloading and implementing a specific model, such as resnet 101, 640x640, retinanet 101, through the tensorflow model zoo, resulting in the direct download of the zip file inside the pre-trained folder.', 'duration': 106.998, 'highlights': ['The process involves efficiently setting up custom object detection by downloading and implementing a specific pre-trained model, such as ResNet 101, 640x640, RetinaNet 101, through the TensorFlow Model Zoo, resulting in the direct download of the zip file inside the pre-trained folder.', 'The speaker demonstrates the process of copying the link address of the chosen pre-trained model and using wget path to download it, leveraging the backend Linux environment in Google Lab.', 'The speaker emphasizes the importance of quickly creating custom object detection and running it for a limited number of times, showcasing the efficiency of the process and providing guidance for the audience to follow the specific steps.', 'The speaker mentions taking a less number of images to demonstrate the quick creation of custom object detection, enhancing the understanding of the process for the audience.']}, {'end': 1360.927, 'start': 1232.307, 'title': 'Extracting and preparing data set for training', 'summary': 'Details the process of downloading and extracting a dataset, creating a label map with two classes (car and bike), and preparing the dataset for training.', 'duration': 128.62, 'highlights': ['The chapter details the process of downloading and extracting a dataset, creating a label map with two classes (car and bike), and preparing the dataset for training.', "The speaker demonstrates the process of downloading and extracting a dataset, highlighting the use of the 'tar xvf' command to extract the tar.zz file.", 'Creating a label map with two classes (car and bike) is explained, with the label map being stored as label_map.pbtxt in the annotation folder.']}], 'duration': 278.792, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z41082135.jpg', 'highlights': ['Setting up custom object detection with pre-trained models like ResNet 101 and RetinaNet 101 through TensorFlow Model Zoo for efficient processing and quick creation of custom object detection.', 'Utilizing Google Colab Pro for longer runtimes and sessions, and leveraging the backend Linux environment for efficient downloading and implementation of pre-trained models.', 'Demonstrating the process of uploading images to the training folder and annotating car with bounding box values, and creating a label map with two classes (car and bike) for dataset preparation.']}, {'end': 2320.005, 'segs': [{'end': 1472.217, 'src': 'heatmap', 'start': 1413.198, 'weight': 0.891, 'content': [{'end': 1416.059, 'text': 'generate underscore tf.record.py.', 'start': 1413.198, 'duration': 2.861}, {'end': 1423.542, 'text': "so if i go over here, you can see that, uh, one of the file that i've actually kept will probably be in the scripts folder.", 'start': 1416.059, 'duration': 7.483}, {'end': 1428.305, 'text': 'So here I have this particular file generated underscore tf record.', 'start': 1424.162, 'duration': 4.143}, {'end': 1433.048, 'text': 'Now where this file I got generated from, I just copied this entire code.', 'start': 1428.865, 'duration': 4.183}, {'end': 1435.09, 'text': "See this entire code, I've just copied it.", 'start': 1433.249, 'duration': 1.841}, {'end': 1439.173, 'text': "Pasted it in a .py file and I've kept that file ready.", 'start': 1436.251, 'duration': 2.922}, {'end': 1445.197, 'text': "Now what I'm going to do, I'm just going to add this particular file inside my training demo folder.", 'start': 1440.053, 'duration': 5.144}, {'end': 1450.361, 'text': "So here I'm just executing and copying it in my training demo folder.", 'start': 1445.797, 'duration': 4.564}, {'end': 1452.702, 'text': 'Now here is my entire file.', 'start': 1451.381, 'duration': 1.321}, {'end': 1454.744, 'text': 'See generate underscore TF record.', 'start': 1452.722, 'duration': 2.022}, {'end': 1458.447, 'text': 'Okay Now what I have to do, I have to just go inside this particular path.', 'start': 1455.224, 'duration': 3.223}, {'end': 1465.392, 'text': "So I'll write CD because I have to execute that particular command, right? CD and execute it.", 'start': 1458.647, 'duration': 6.745}, {'end': 1466.913, 'text': "Now I'm inside my training demo.", 'start': 1465.612, 'duration': 1.301}, {'end': 1472.217, 'text': 'Now in order to execute that particular file, I just have to copy this entire command.', 'start': 1467.893, 'duration': 4.324}], 'summary': 'The speaker describes generating a tf.record.py file and copying it to a training demo folder.', 'duration': 59.019, 'max_score': 1413.198, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z41413198.jpg'}, {'end': 1817.38, 'src': 'heatmap', 'start': 1776.108, 'weight': 2, 'content': [{'end': 1778.949, 'text': 'Then the next line will be basically this fine-tuned checkpoints.', 'start': 1776.108, 'duration': 2.841}, {'end': 1785.172, 'text': 'In order to get this fine-tuned checkpoints, I have to basically take this particular path from my..', 'start': 1779.57, 'duration': 5.602}, {'end': 1787.072, 'text': 'pre-trained model.', 'start': 1786.252, 'duration': 0.82}, {'end': 1790.774, 'text': "so let's go over here, and this will be my real real train model.", 'start': 1787.072, 'duration': 3.702}, {'end': 1792.315, 'text': "right from this i'm going to refer it.", 'start': 1790.774, 'duration': 1.541}, {'end': 1797.437, 'text': "so i'm just going to go over here, go into my checkpoint and then i'm just going to copy this entire path.", 'start': 1792.315, 'duration': 5.122}, {'end': 1802.699, 'text': 'okay, so my checkpoint probably is in line number 161.', 'start': 1797.437, 'duration': 5.262}, {'end': 1806.601, 'text': "okay, so this 161, i'll just go and update it.", 'start': 1802.699, 'duration': 3.902}, {'end': 1807.861, 'text': 'so here is my 161.', 'start': 1806.601, 'duration': 1.26}, {'end': 1814.299, 'text': 'okay, I am pasting it.', 'start': 1807.861, 'duration': 6.438}, {'end': 1816.12, 'text': 'I will just remove this index.', 'start': 1814.319, 'duration': 1.801}, {'end': 1817.38, 'text': "I don't require this index.", 'start': 1816.16, 'duration': 1.22}], 'summary': 'Updating checkpoint path to line 161, removing index.', 'duration': 41.272, 'max_score': 1776.108, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z41776108.jpg'}, {'end': 2067.87, 'src': 'heatmap', 'start': 2035.593, 'weight': 1, 'content': [{'end': 2036.894, 'text': 'now what all path i need to change.', 'start': 2035.593, 'duration': 1.301}, {'end': 2038.314, 'text': 'first of all, my model directory.', 'start': 2036.894, 'duration': 1.42}, {'end': 2041.756, 'text': 'model directory is nothing but this, this specific directory.', 'start': 2038.314, 'duration': 3.442}, {'end': 2045.337, 'text': "so i'm just going to remove this and paste it over here.", 'start': 2041.756, 'duration': 3.581}, {'end': 2052.82, 'text': "okay, similarly, i'm just going to remove this path and paste it over here with the pipeline config.", 'start': 2045.337, 'duration': 7.483}, {'end': 2055.241, 'text': 'okay, only these two changes will be there.', 'start': 2052.82, 'duration': 2.421}, {'end': 2056.541, 'text': 'fine, uh, we have done this.', 'start': 2055.241, 'duration': 1.3}, {'end': 2057.842, 'text': "i'll go over here and execute it.", 'start': 2056.541, 'duration': 1.301}, {'end': 2062.706, 'text': "Now. once I execute it, you'll be able to see that probably the training will start.", 'start': 2058.402, 'duration': 4.304}, {'end': 2067.87, 'text': "I'm getting some errors saying that no file or such directory.", 'start': 2064.306, 'duration': 3.564}], 'summary': 'Making changes to model directory and pipeline config to resolve errors in training process.', 'duration': 32.277, 'max_score': 2035.593, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42035593.jpg'}, {'end': 2189.399, 'src': 'embed', 'start': 2164.065, 'weight': 1, 'content': [{'end': 2169.348, 'text': 'Inside my pipeline config, I want to see that how many number of step size that I keep it.', 'start': 2164.065, 'duration': 5.283}, {'end': 2174.571, 'text': "So again, number of step size I've kept it as, total steps I've kept it as 25,000.", 'start': 2169.808, 'duration': 4.763}, {'end': 2175.932, 'text': "It's obviously going to take time.", 'start': 2174.571, 'duration': 1.361}, {'end': 2180.434, 'text': "So what I'm going to do is that I'm just going to make it as 2,000, you know, 2,000 step size.", 'start': 2176.432, 'duration': 4.002}, {'end': 2184.897, 'text': 'Make sure that you make the changes in 152 and 162 like, okay, quickly.', 'start': 2180.974, 'duration': 3.923}, {'end': 2189.399, 'text': "So I'm going to save it, close this, and again, execute it once again.", 'start': 2185.337, 'duration': 4.062}], 'summary': 'Pipeline config: reducing total steps from 25,000 to 2,000 to save time and making changes in 152 and 162.', 'duration': 25.334, 'max_score': 2164.065, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42164065.jpg'}, {'end': 2320.005, 'src': 'embed', 'start': 2292.928, 'weight': 0, 'content': [{'end': 2299.633, 'text': "So this type, if you're getting it, that basically means that you are, your training is happening properly.", 'start': 2292.928, 'duration': 6.705}, {'end': 2306.329, 'text': 'Okay So guys, the training has happened.', 'start': 2299.953, 'duration': 6.376}, {'end': 2314.819, 'text': 'It has completed 2000 steps and you can see the loss function initially was 17.149 and it is coming till 10.', 'start': 2306.409, 'duration': 8.41}, {'end': 2320.005, 'text': "Just to show you quickly, I've stopped the training right now, but let's see how the performance of the specific model is there.", 'start': 2314.819, 'duration': 5.186}], 'summary': 'Training completed 2000 steps with loss decreasing from 17.149 to 10.', 'duration': 27.077, 'max_score': 2292.928, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42292928.jpg'}], 'start': 1360.927, 'title': 'Preparing and training object detection model', 'summary': 'Covers converting xml files to tfrecord format, configuring the training pipeline for object detection, and model training involving 2000 steps with a reduction in the loss function from 17.149 to 10.', 'chapters': [{'end': 1601.956, 'start': 1360.927, 'title': 'Convert xml to tfrecord', 'summary': 'Details the process of converting xml files with bounding boxes into tfrecord format, involving the execution of a python script to generate the tfrecord files for both the training and test datasets.', 'duration': 241.029, 'highlights': ['The process involves running a Python script called generate_tf_record.py to convert the XML files into TFRecord format for the training and test datasets. The Python script generate_tf_record.py is executed to convert the XML files into TFRecord format for both the training and test datasets.', 'The specific steps for executing the script involve providing the paths for the images folder, label_map.pbtxt, and the output TFRecord files for both the training and test datasets. The execution of the script requires providing the paths for the images folder, label_map.pbtxt, and the output TFRecord files for the training and test datasets.', 'The process demands patience and attention to detail in following the specific steps to ensure the successful conversion of the XML files into TFRecord format. Patience and attention to detail are necessary to successfully follow the specific steps for converting the XML files into TFRecord format.']}, {'end': 1946.127, 'start': 1602.357, 'title': 'Configuring training pipeline for object detection', 'summary': 'Involves configuring the training pipeline for object detection using a pre-trained model, including creating folders, updating pipeline config, and making necessary changes such as the number of classes, batch size, checkpoints, label map paths, and input paths.', 'duration': 343.77, 'highlights': ['The first change to be made is setting the number of classes to 2, followed by updating batch size to 8 for limited RAM usage.', 'Updating fine-tuned checkpoints path from the pre-trained model, along with setting detection instead of classification and configuring label map paths for train and test data.', 'Configuring input paths for train and test data, ensuring the necessary changes are made for the training pipeline to function effectively.']}, {'end': 2320.005, 'start': 1946.167, 'title': 'Pipeline configuration and model training', 'summary': 'Discusses updating the pipeline configuration, making changes to the model directory and pipeline config, and executing the code for training the model, which completes 2000 steps with a reduction in the loss function from 17.149 to 10.', 'duration': 373.838, 'highlights': ['The model completes 2000 steps with a reduction in the loss function from 17.149 to 10, indicating successful training. The training process completes 2000 steps with the loss function decreasing from 17.149 to 10, demonstrating successful model training.', 'The pipeline configuration is updated with specific changes made in the step size to ensure efficient execution, with the total steps adjusted to 2000. The pipeline configuration is modified to set the total steps to 2000, aiming for efficient execution of the training process.', 'The process involves making changes to the model directory and pipeline config, ensuring the correct paths are specified for seamless execution. Changes are made to the model directory and pipeline config, ensuring accurate path specifications for smooth execution of the training process.']}], 'duration': 959.078, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z41360927.jpg', 'highlights': ['The training process completes 2000 steps with the loss function decreasing from 17.149 to 10, demonstrating successful model training.', 'The pipeline configuration is modified to set the total steps to 2000, aiming for efficient execution of the training process.', 'Updating fine-tuned checkpoints path from the pre-trained model, along with setting detection instead of classification and configuring label map paths for train and test data.']}, {'end': 2826.521, 'segs': [{'end': 2426.75, 'src': 'embed', 'start': 2401.889, 'weight': 0, 'content': [{'end': 2407.773, 'text': 'now, if i go to my pipeline config path, this should be the path it should get updated with, right.', 'start': 2401.889, 'duration': 5.884}, {'end': 2409.634, 'text': "so i'm just going to copy this particular path.", 'start': 2407.773, 'duration': 1.861}, {'end': 2416.921, 'text': "Okay, and then I'm just going to remove this, because every command you need to change this specific path.", 'start': 2410.114, 'duration': 6.807}, {'end': 2419.023, 'text': 'guys. then my train checkpoint.', 'start': 2416.921, 'duration': 2.102}, {'end': 2422.766, 'text': 'again, this should be my my directory path.', 'start': 2419.023, 'duration': 3.743}, {'end': 2425.109, 'text': 'So this is my directory path over here.', 'start': 2423.287, 'duration': 1.822}, {'end': 2426.75, 'text': "I'm just going to change it.", 'start': 2425.729, 'duration': 1.021}], 'summary': 'Updating pipeline config and directory paths for commands.', 'duration': 24.861, 'max_score': 2401.889, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42401889.jpg'}, {'end': 2501.081, 'src': 'embed', 'start': 2472.736, 'weight': 1, 'content': [{'end': 2477.319, 'text': "you'll be able to see a PB file that will be present inside it.", 'start': 2472.736, 'duration': 4.583}, {'end': 2479.54, 'text': 'okay?. We are exporting the model.', 'start': 2477.319, 'duration': 2.221}, {'end': 2482.642, 'text': 'exporting the model from here right? From this particular checkpoint right?', 'start': 2479.54, 'duration': 3.102}, {'end': 2485.444, 'text': 'Because here we have this entire train checkpoints.', 'start': 2483.002, 'duration': 2.442}, {'end': 2487.425, 'text': 'So again, it will take some time.', 'start': 2486.264, 'duration': 1.161}, {'end': 2490.147, 'text': "Let's see how much time it will probably take.", 'start': 2488.686, 'duration': 1.461}, {'end': 2496.019, 'text': 'skipping, full serialization, all these things, kind of things are there.', 'start': 2493.078, 'duration': 2.941}, {'end': 2497.72, 'text': 'so here you will be able to see.', 'start': 2496.019, 'duration': 1.701}, {'end': 2501.081, 'text': 'if you are getting this kind of error, then probably you have done something wrong.', 'start': 2497.72, 'duration': 3.361}], 'summary': 'Exporting model from train checkpoints. full serialization may take some time.', 'duration': 28.345, 'max_score': 2472.736, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42472736.jpg'}, {'end': 2683.479, 'src': 'embed', 'start': 2638.607, 'weight': 4, 'content': [{'end': 2644.93, 'text': 'annotation path perfect, uh, and then i just have to make that many number of changes and we are good to go.', 'start': 2638.607, 'duration': 6.323}, {'end': 2649.532, 'text': "now i'll just try to execute it and let's see how will i get an accuracy.", 'start': 2644.93, 'duration': 4.602}, {'end': 2651.413, 'text': "so it's loading a model.", 'start': 2649.532, 'duration': 1.881}, {'end': 2659.457, 'text': "so what i've done is that i've given this specific image you know this 18th image and probably it will try to do an object detection.", 'start': 2651.413, 'duration': 8.044}, {'end': 2663.819, 'text': 'so this specific image is there and it will try to do an object detection for this.', 'start': 2659.457, 'duration': 4.362}, {'end': 2666.74, 'text': "okay, so let's try, let's try it out.", 'start': 2663.819, 'duration': 2.921}, {'end': 2674.235, 'text': 'So, running inferences for this particular 18, image 18, probably you may not get a good result also,', 'start': 2669.133, 'duration': 5.102}, {'end': 2676.676, 'text': 'but at least you should be getting some kind of bounding boxes.', 'start': 2674.235, 'duration': 2.441}, {'end': 2681.278, 'text': 'So no, the bounding boxes has not come.', 'start': 2677.876, 'duration': 3.402}, {'end': 2683.479, 'text': 'Let me try some other images if possible.', 'start': 2681.458, 'duration': 2.021}], 'summary': 'Testing object detection model on image 18 for bounding boxes.', 'duration': 44.872, 'max_score': 2638.607, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42638607.jpg'}, {'end': 2815.559, 'src': 'embed', 'start': 2775.069, 'weight': 3, 'content': [{'end': 2781.611, 'text': "Now the thing is that if I keep on increasing the number of training steps, probably I'll try to make it to 25,000,", 'start': 2775.069, 'duration': 6.542}, {'end': 2784.672, 'text': "then you'll be able to precisely get the output like this.", 'start': 2781.611, 'duration': 3.061}, {'end': 2786.512, 'text': 'okay?. And this again.', 'start': 2784.672, 'duration': 1.84}, {'end': 2791.553, 'text': 'after that 2,500, I again retrain it for another 2,500 steps and it was getting improved.', 'start': 2786.512, 'duration': 5.041}, {'end': 2796.135, 'text': "And I will still try more for more number of steps and then probably I'll try to check it.", 'start': 2792.154, 'duration': 3.981}, {'end': 2800.976, 'text': 'But again, the main agenda was this to show you how you can actually do a custom object detection.', 'start': 2796.175, 'duration': 4.801}, {'end': 2804.197, 'text': "right. so don't worry about all the files.", 'start': 2801.596, 'duration': 2.601}, {'end': 2812.018, 'text': "i will definitely make sure that i'll give you all the files in a zip, probably by uploading in my google drive, and you can just follow the steps.", 'start': 2804.197, 'duration': 7.821}, {'end': 2814.179, 'text': 'make sure that you follow the documentation also.', 'start': 2812.018, 'duration': 2.161}, {'end': 2815.559, 'text': 'so i hope you like this particular video.', 'start': 2814.179, 'duration': 1.38}], 'summary': 'Training steps increased to 25,000, showing improved results in custom object detection.', 'duration': 40.49, 'max_score': 2775.069, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42775069.jpg'}], 'start': 2320.606, 'title': 'Exporting trained model and custom object detection demo', 'summary': 'Explains the process of exporting a trained object detection model, including copying and pasting specific code, updating paths, creating and changing directories, and the expected outcome of exporting the model as a pb file, with some time taken for the process. it also demonstrates the process of custom object detection, showing the training and inference process, achieving improved results by increasing training steps, and offering to provide necessary files for replication.', 'chapters': [{'end': 2496.019, 'start': 2320.606, 'title': 'Exporting trained model', 'summary': 'Explains the process of exporting a trained object detection model, including copying and pasting specific code, updating paths, creating and changing directories, and the expected outcome of exporting the model as a pb file, with some time taken for the process.', 'duration': 175.413, 'highlights': ['The process involves copying and pasting specific code, updating paths, and creating and changing directories. None', 'The expected outcome of the process is the generation of a PB file containing the exported model. None', 'The chapter mentions that the process may take some time to complete. None']}, {'end': 2826.521, 'start': 2496.019, 'title': 'Custom object detection demo', 'summary': 'Demonstrates the process of custom object detection, showing the training and inference process, achieving improved results by increasing training steps, and offering to provide necessary files for replication.', 'duration': 330.502, 'highlights': ["By increasing the training steps from 2500 to 25000, the accuracy of the custom object detection improved, demonstrating the impact of training duration on the model's performance.", "The process of custom object detection involves running inferences on specific images and evaluating the accuracy by observing the bounding boxes, with the goal of enhancing the model's performance through additional training.", 'The presenter emphasizes the importance of training the model multiple times to detect and rectify errors, highlighting the iterative nature of refining the model for improved performance and accuracy.', 'The presenter offers to provide all necessary files for replication and encourages viewers to subscribe for future content, demonstrating the intention to engage with the audience and provide ongoing support.']}], 'duration': 505.915, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/XoMiveY_1Z4/pics/XoMiveY_1Z42320606.jpg', 'highlights': ['The process involves copying and pasting specific code, updating paths, and creating and changing directories.', 'The expected outcome of the process is the generation of a PB file containing the exported model.', 'The chapter mentions that the process may take some time to complete.', "By increasing the training steps from 2500 to 25000, the accuracy of the custom object detection improved, demonstrating the impact of training duration on the model's performance.", "The process of custom object detection involves running inferences on specific images and evaluating the accuracy by observing the bounding boxes, with the goal of enhancing the model's performance through additional training.", 'The presenter emphasizes the importance of training the model multiple times to detect and rectify errors, highlighting the iterative nature of refining the model for improved performance and accuracy.', 'The presenter offers to provide all necessary files for replication and encourages viewers to subscribe for future content, demonstrating the intention to engage with the audience and provide ongoing support.']}], 'highlights': ['The training process completes 2000 steps with the loss function decreasing from 17.149 to 10, demonstrating successful model training.', "By increasing the training steps from 2500 to 25000, the accuracy of the custom object detection improved, demonstrating the impact of training duration on the model's performance.", 'Setting up custom object detection with pre-trained models like ResNet 101 and RetinaNet 101 through TensorFlow Model Zoo for efficient processing and quick creation of custom object detection.', 'The presenter emphasizes the importance of training the model multiple times to detect and rectify errors, highlighting the iterative nature of refining the model for improved performance and accuracy.', 'The process involves copying and pasting specific code, updating paths, and creating and changing directories.']}