Train a Custom YOLOv5 Model on Google Colab and Detect Objects In Real Time on Windows | Part 3

🚀Hey everyone and welcome to the final part of the YOLOv5 series!

In this video (Part 3) we'll learn about training a custom YOLOv5 PyTorch model on Google Colab, and using Weights & Biases to understand the training process.

We'll start off by discussing the differences between pre-trained and custom-trained models, and how we can train using the custom dataset that we collected in part 2. Then, we'll use Weights & Biases to determine which of the models we trained is the best one, and deploy that model locally on Windows to perform real-time detection using a custom-trained model.

---

Links

---

You can watch the whole YOLOv5 Series here:

---

Follow Ivan:

---

⏳ Timestamps ⏳
00:00 Intro
00:50 How to think about training a YOLOv5 to detect new classes
03:28 Training a custom YOLOv5 model on Google Colab
08:23 How Weights & Biases helps you understand the training process
12:12 Training a large YOLOv5 model
16:18 Deploying a custom-trained model on Windows
18:28 Outro

[WeightsBiases]

Thank you for watching part 3 of the YOLOv5 Series!

You can watch the whole YOLOv5 Series here:

[aryashinde]

thank you so much for your help. i was able to load my custom YOLOv5 model on windows and it works perfectly. Im extremely grateful to you!

[bobsalita]

I had no idea that W&B was making such informative and quirky videos. Really enjoyed this and will backprop others.

[singgihbekti]

hi, I have some confusions, how to make data logging for detected object to CSV or XLSX format? then we can see the number of detected object according to specific classes? thank you

[ajohnny]

The whole series helped me a lot! Very appreciate

[lamkitwong]

I encounter this: AttributeError: Can't get attribute 'DetectionModel' on <module 'models.yolo' from
Anyone know how to solve this error?

[adarshs]

I get the error "notimplementederror cannot instantiate posixpath on your system yolov5". I did everything as per the video. I dunno how i get this error and he doesn't. pls tell me a solution to resolve this issue

[krixz]

thank you so much for the effort bro u earned my respect and subscription

[lalithchakradhar]

Hey, what is the command to use WiFi connected camera for object detection?

[lightning_best]

Hello. I was inspired by your videos and decided to study yolov5. But I ran into one problem: there is a person box, 'person' is written above the box, but this word is filled with a red background . I want to make sure there is no red padding behind the text. Could you suggest what needs to be changed in the code to fix this problem. I will be very grateful to you
P.S. I opened the plots.py file, but there I could only adjust the text size and box

[ozcankosar]

Can I use models I trained on windows for ROS on ubuntu?

[arekdevss]

I followed your tutorial and at the last step when running train.py I get this error mesage: ValueError: Instances of wandb.Artifact and wandb.apis.public.Artifact can only be top level keys in wandb.config, everything worked until that point and I cant find anything online about this error

[moahaimen]

hi, I did train the custom_data model in yolov5 using Google Colab Pro. And store the trained model using "Pickle" I need now to load it and work on it to detect the test images I suppose that should save me hours of training, How can i do that please

[robertdaniels]

I am interested in AI by watching person of interest, do you think ASI is possible someday?
Also your channel is amazing and videos are very helpful👍

[aliff]

Hi, thank you for your video. It helps me very much in YOLOv5. I have a question. How do I choose my detect.py to be the model that I train? Or is it automatically chosen?

I'm new so I hope you can help me

[snehitvaddi]

Btw, does OpenCV supports working with YoloV5 models ?

[ruchikaarora]

Hello sir,
How we can deploy the model on static images only

[je]

Love the videos! When I'm trying to train the model I get the error: TypeError: Class labels must be a dictionary of numbers to string. I have tried updating python wandb however, this creates another error which was solved in the comments of reverting back to the previous version of Wandb. I was wondering if you had any idea how to fix this?

[lightning_best]

Hello. I was inspired by your videos and decided to study yolov5. But I ran into one problem: there is a person box, 'person' is written above the box, but this word is filled with a red background . I want to make sure there is no red padding behind the text. Could you suggest what needs to be changed in the code to fix this problem. I will be very grateful to you
P.S. I opened the plots.py file, but there I could only adjust the text size and box