Train Your Custom YOLO Models In A Few Clicks with Ultralytics HUB | Episode 10

In this episode, you will learn about the Ultralytics HUB and how easy it is to take your dataset, set up and model with hyperparameters, and then train the Ultralytics YOLOv8 model without writing any code at all. We are also going to go over all the possibilities and options with the Hub and even show you real-time object detection with YOLOv8 on a phone. You can try it out on both Android and IOS.
Learn more ➡️ www.ultralytics.com/blog/how-...
🔗 Colab Notebook: colab.research.google.com/git...
Ultralytics HUB on Apple Store: bit.ly/3HPAw4T
Ultralytics HUB on Google Play Store: bit.ly/3Ld80LC
🌟NEW: Register for YOLO VISION 2024: yolovision.ultralytics.com/
Ultralytics ⚡ resources
- About Us - ultralytics.com/about
- Join Our Team - ultralytics.com/work
- Contact Us - ultralytics.com/contact
- Discord - ultralytics.com/discord
- Ultralytics License - ultralytics.com/license
YOLOv8 🚀 resources
- GitHub - github.com/ultralytics/ultral...
- Docs - docs.ultralytics.com/

[@WavaDev]

Is it possible to further train the model with additional data without restarting the training process? After training the model for 4 hours, I prefer not to start over.

[@Ultralytics]

If you intend to train the model with extra data, simply utilize the last saved model file. For instance, if you have completed training for 50 epochs and obtained the `best.pt` file, you can subsequently employ this model file for further training with additional data.
```yolo train model="path/to/best.pt" data="path/to/data.yaml" epochs=10, ...```
Thanks
Ultralytics Team!

[@micuentadecasa]

Congrats for your work. I visited Ultralytics hub, and I have a question, what is the difference between Train with agents vs Train Models with Ultralytics Cloud?

[@DanielGonzalez-wr7fz]

First refers to train with your local machine, and Ultralytics Cloud is coming soon.

[@Ultralytics]

Hi there! Thank you for your kind words. Training with agents entails the process of training the model on your local system, with the capability to view training graphs and metrics on Ultralytics HUB. On the other hand, Ultralytics Cloud offers the option to train the model in a cloud-based environment.

[@ripunjaysingh8003]

Is it possible to get best.pt file once the training gets completed on Ultralytics HUB???

[@Ultralytics]

Once training is complete, you can obtain the best.pt file by clicking on Deploy and choosing the PyTorch format. This will generate the best.pt file, which you can readily use for your specific task.

[@m033372]

How does the Ultralytics HUB handle imbalanced datasets, and are there any built-in features to help ensure accurate model training in such scenarios?

[@Ultralytics]

Great question! Ultralytics HUB offers several features to help manage imbalanced datasets. You can use data augmentation techniques and class weighting to ensure your model trains accurately. Additionally, the platform provides detailed analytics and visualizations to help you monitor class distribution and model performance. Make sure you're using the latest versions of our tools for the best experience. For more details, check out our documentation at docs.ultralytics.com/hub/quickstart/. If you have any specific issues or need further assistance, feel free to share more details! 😊

[@giuliaghinassi1734]

Hi, I'm a complete newbie and I tried to follow your instruction BUT when I run the "hub login" a warning pops up:
WARNING ⚠ Unable to automatically guess model task, assuming 'task=detect'. Explicitly define task for your model, i.e. 'task=detect', 'segment', 'classify','pose' or 'obb'."
And honestly I do not understand what is going on because it seems that in your case everything goes perfectly smooth🥵

[@Ultralytics]

It appears that you're encountering a warning message when running "hub login," indicating that the model task couldn't be automatically determined. To resolve this, specify the task explicitly by adding 'task=detect', 'segment', 'classify', 'pose', or 'obb' to your command. This step ensures proper model configuration.

[@imadahmad104]

Ultralytics hub is not accessible, please sort out. not able to login and register

[@Ultralytics]

Apologies for any inconvenience. Could you kindly provide the error logs you encountered? We strongly recommend submitting Ultralytics Hub issues to our Ultralytics Hub's Issue section for a more prompt and effective response.
github.com/ultralytics/hub/issues

[@imadahmad104]

@@UltralyticsFirst it was not able to login now i for model training its not going to step 3 and stuck at waiting for connection

[@Ultralytics]

@imadahmad104, the problem should be resolved now. Our team has successfully addressed the issues that users were encountering during the signup process yesterday.
Thanks

[@user-mh8kb5ng9f]

Hi, Ultralytics HUB is really convenient with concise GUI. But I have trouble when training. When I use Google Colab to train YOLOv8n model with Simpon dataset, Google Colab lose connection with Ultralytics HUB after the first epoch. Warning says "Your dialog has crashed and will restart automatically" in Google Colab. I trained using YOLOv5 returned the same result

[@Ultralytics]

It seems there's a disconnection issue during YOLOv8n training on Google Colab with Ultralytics HUB. Check for Colab session timeouts, resource availability, and data/model compatibility. Saving checkpoints more frequently may help resume training after disconnections.
Thanks