Tensorflow Object Detection in 5 Hours with Python | Full Course with 3 Projects

Want to get up to speed on AI powered Object Detection but not sure where to start?

Want to start building your own deep learning Object Detection models?

Need some help detecting stuff for your course, startup or business?

This is the course you need!

In this course, you’ll learn everything you need to know to go from beginner to practitioner when it comes to deep learning object detection with Tensorflow. This course mainly revolves around Python but there’s a little Javascript thrown in as well when it comes to building a web app in Project 2. But don’t fret we’ll take it step by step so you can take your time and work through it. All the code it made available through GitHub, links below.

As part of this course you’ll build four different object detection models:
A. Gesture Detection - this is the first project where you’ll be able to build a model that detects four different gestures
B. Microscope Based Defect Detection - here we’ll leverage a USB microscope to detect defects in LEDs and PCBs using TFOD and Python
C. Web Direction Detection - in this model you’ll learn how to detect hand directions for integration in a React Js Web App with Tensorflow Js
D. Face Sentiment Detection - here you’ll learn how to estimate facial sentiment using Tensorflow Object Detection on a Raspberry Pi with TFLite

You’ll learn how to:
1. Install Tensorflow Object Detection on a Local Machine and on Colab
2. Collect and Label images for Object Detection using LabelImg
3. Train Deep Learning powered Object Detection Models using Python and TFOD
4. Detect objects in real time using a webcam and using Images
5. Tune Object Detection models to improve Precision and Recall
6. Export your model to Tensorflow JS for integration in React JS web apps
7. Export your model to TFLite for use on a Raspberry Pi

Get the code

Chapters:
0:00 - Start
12:13 - SECTION 1: Installation and Setup
26:34 - Cloning the Baseline Code from GitHub
27:59 - Creating a Virtual Environment
39:57 - SECTION 2: Collecting Images and Labelling
44:48 - Collecting Images Using Your Webcam
1:04:11 - Labelling Images for Object Detection using LabelImg
1:29:08 - SECTION 3: Training Tensorflow Object Detection Models
1:34:04 - Tensorflow Model Zoo
1:39:04 - Installing Tensorflow Object Detection for Python
1:56:41 - Installing CUDA and cuDNN
2:06:42 - Using Tensorflow Model Zoo models
2:09:21 - Creating and Updating a Label Map
2:10:09 - Creating TF Records
2:17:23 - Training Tensorflow Object Detection Models for Python
2:27:48 - Evaluating OD Models (Precision and Recall)
2:29:08 - Evaluating OD Models using Tensorboard
2:34:07 - SECTION 4: Detecting Objects from Images and Webcams
2:34:52 - Detecting Objects in Images
2:38:57 - Detecting Objects in Real Time using a Webcam
2:41:56 - SECTION 5: Freezing TFOD and Converting to TFJS and TFLite
2:42:25 - Freezing the Tensorflow Graph
2:44:17 - Converting Object Detection Models to Tensorflow Js
2:45:27 - Converting Object Detection Models to TFLite
2:47:45 - SECTION 6: Performance Tuning to Improve Precision and Recall
3:12:34 - SECTION 7: Training Object Detection Models on Colab
3:24:05 - SECTION 8: Object Detection Projects with Python
3:25:25 - Project 1: Detecting Object Defects with a Microscope
3:57:34 - Project 2: Web Direction Detection using Tensorflow JS
4:47:40 - Project 3: Sentiment Detection on a Raspberry Pi Using TFLite

Oh, and don't forget to connect with me!

Happy coding!
Nick

P.s. Let me know how you go and drop a comment if you need a hand!

[mrx-_-k]

Hey man, just wanted to tell you despite the fact that this video was published 2 years ago it's still standing the test of time and that means a lot of people will appreciate you because of your cool videos.keep it up

[GeorgeTrialonis]

Thank you for being so specific and explicit. This counts a lot especially for people with little experience in AI. Every minor detail counts and saves one a lot of frustration. Thank you.

[pedrolima-eupb]

i have watched 5 mins of this and i am addicted. As a person with ADHD i really struggle to find tutorials that are stimulating enough for me and this scratches all the right itches. i know this takes a lot of work to make and i thank you so much for it.

[Death_User]

This is the only channel I found so far where the person isn't just talking out of their ass about AI this guy is actually doing it and proving his skill and teaching it to others you sir are a legend

[DhruvPatel-mgou]

Making long length video requires lot of efforts and you deserve my like and love your content as always😊😊

[markd]

This video is amazing, the best image detection guide I’ve seen. Would love to see a video on auto-labeling, but even without that this channel is a must subscribe.

[NeverForgetNasa]

What an insane amount of work this video must have required! This course is just amazing! You deserve my full respect.

[manikanthgoud]

Absolutely love the way you go through the whole thing w us! Only a handful of people do this and Hats off to you! You are single handedly carrying millions of students on your back :D

[meowmaple]

This video got me into successfully training my first tensorflow object detection model! Having previously taken an andrew ng course on ML, I was looking for tensorflow tutorials online but they were all quite messy and hard to set up. Im glad I followed through your video which was straightforward and extremely helpful! Can't thank you enough man :))

[haltersweb]

Your fabulous effort at making this comprehensive video is to be commended. You have a true teacher’s spirit. Thank you for all you do, and for how much you care.

[noohagaad]

The best tutorial I have ever seen in my LIFE
Not only in object detection, but also in Python

[datapro]

I keep coming back to this video in my day-to-day work. It is invaluable. Thank you Nick!

[JasonMann]

Hi Nicholas, thank you so much for your tutorial! You got me off the ground & started on my way to counting individual electronics components for my work project.

For those who are getting a "

[flynn]

Hi Nick, the amount of detail you provide is what we need, especially for the more complex stuff like knowing the different TF models and where to find them. I've been watching your stuff for a while now, but wish I had started to watch this course sooner! Sure it's long, but at least you put it into chapters making it easier for you to produce and us to watch.

[briancarroll]

Thank you, Nicholas, for this thorough course. I appreciate the amount of time it took to make this. I had trouble with the installation and couldn't get either the image or video detection to work. I decided to do it all over again and it worked. There were 2 errors that occurred: module not found for 'official' and for 'object_detection'. This is on a Windows machine. I ultimately solved it by copying the TensorFlow\models\official and folders into the tfod\lib\site_packages folder.

[ricarprieto]

two words... LOVE YOU

thank you, I recently discovered your channel and it's amazing!!

[pedropintogarcia]

This was the best Tensorflow course I've seen so far. Everyone starting out can learn a lot from this, I definitely did
Thanks a lot for making up the effort to put all this together, wish you all the best!!!

[jubaerhasan]

Sir, you have no idea how you saved me a thousand of hours trying to learn and look for a full on course that is actually beginner friendly <3 much respect, Subbed. <3

[danielleivy]

I'm only an hour in and this is by far the best course I've seen. It's practical and goes through real world customizable code step by step. This is exactly what I needed after wasting hours on other courses. After 3B1B and Stanford CS229 math, I needed a practical way and this is it.👍🏾
Edit: 3B1B and CS229 are not the wasted courses I was talking about (I love Grant and Ng). They're very much needed theory, but so many other 'practical' courses out there don't apply it to actual projects..

[lzqdtsm]

I went through this video second by second and I can only say your video and tutorials are VERY impressive and easy to follow along! Detailed explanations of what your codes are actually doing allowed me to understand how object detection works and apply new ideas (exactly how transferred learning works xD). Thanks a million, and I hope I can see more of your awesome tutorials!