How to Train custom Object Detection Neural Network using TensorFlow 2.3 on Google Colab Free GPU

#tensorflow #objectdetection #computervision
If you like the video, please subscribe to the channel by using the below link

Link for my deeplearning udemy course coupon code added

Hi Everyone, In this tensorflow tutorial, I have explained how to train tensorflow object detection api with your own data. I have used tensorflow latest version here. For

model training, I am using google colab free GPU. It means it’s a full-on tutorial on how to train object detection api with your own data or custom object detection model on google colab.
I have taken the example of playing card detection in this video and the model is taken from tensorflow model zoo.

-----Time links to each step in the video-----

Chapters:
0:00 Introduction
3:24 Install TensorFlow object detection api using powershell,Set up Object Detection directory and python virtual environment at once
6:30 Gather and label pictures
13:51 Generate training and testing dataset
16:39 Create train tfrecord and test tfrecord files
20:23 Create label map and configure training
28:55 Setup google colab for object detection model training
33:20 Start model training on colab
33:56 Export inference graph
36:57 Try out your object detector for images.
39:46 Try out your object detector on a live webcam.

this contains:
a. PowerShell script to install and set up tensorflow object detection api.
3. run this command for generating csv file for training and testing images
5. Generate tfrecord file for training by this command
6. Generate tfrecord file for training by this command
7. Here are the argument to be updated on the config file for model training
num_classes: 5 [give number of classes here]
learning_rate_base: 0.8e-3
warmup_learning_rate: 0.0001
fine_tune_checkpoint: "efficientdet_d0_coco17_tpu-32/checkpoint/ckpt-0"
fine_tune_checkpoint_type: "detection"

8. train model command

9. export infrence graph command

[ashwinbabu]

i have been searching for good object detection video for so long.
its really great, I loved it. thankyou

[zanuarer]

It is helped me a lot, Thank you very much!
I trained using faster-rcnn and tensorflow 2.

[susantasharma]

One of the complete video on this topic. Thanks for sharing.

[MahmoudMahmoudTobi]

I can tell you had a lot of fun making this. Thanks for the tutorial and the easy setup file!.

[gautamkinfo]

the power sell is does not working at my system. it's open power sell window and then crash...

[JemmyWu]

Awesome! It's indeed the best guide for learning Colab TF2. Thank you so much!

[Sandeepsharma-mcer]

After trying many tutorials this tutorial helps me lot !!
Nicely explained!!

[francescovender]

Hi, can i know what is the problem with this? When installing the Visual C++, it keeps telling me that I have missing packages or damaged packages for setup. I clicked either option for downloading the packages but to no avail. Can you help me on this?

[rahulitsrahull]

Hello Jay.... this is a great tutorial. If you could upload a video on how to convert the saved pb file into TFlite model, it would be much helpful. Thanks

[anilkoluacik]

setup file is either missing or damaged while installing vs c++ build tools. Pls help me to handle this issue.

[kenzilamberto]

Your videos are best at explaining, thank you brother

[robinrai]

thanks for amazing tutorial sir i trained the image up to 9000 steps with minimum loses but not detecing the image it just show the test images

[mr.millenniamachine]

Thank you. How to improve detection time to about 0.1 sec per frame? Please help.

[johnurena]

i get an error when trying to install visual studio ( a setup package is either missing or damaged)

[saidmaroun]

thank you very your explanations, the video was so great and well presented. keep up the great work

[pavankumarpotta]

I am setup the environment
The visual studio package is not getting installed it is throwing me an error

[nicolasachury]

First of all, thanks for the amazing tutorial!!! I would like to ask you the following: I trained my model using Google Colab and I interrupted the processes after 30 min. Now the results are not what I was expecting, so I want to train the model much longer. How can I resume the training process using the checkpoints? any insight from you would be highly appreciated!

[annie]

Hi, can you please guide me on how can I find the confusion matrix, if I am using ssd_mobnet model of object detection API?

[seharsana]

Can you please guide me...i am facing error IMPORTERROR:cannot import name 'model_lib_v2' from object_detection

[malikgaming]

self._read_buf =
: The filename, directory name, or volume label syntax is
; no protocol option
facing this error please help