augment vs auto_augment

[Zvyozdo4ka]

what is the difference between these hyperparameters? augment vs auto_augment in Yolov11? I checked this page, it only mentions only auto_augment, but in Yolov8 it presents

[UltralyticsAssistant]

👋 Hello @Zvyozdo4ka, thank you for your interest in Ultralytics 🚀! We recommend a visit to the Docs for new users where you can find many Python and CLI usage examples and where many of the most common questions may already be answered.

It seems like you are exploring augmentation hyperparameters 🤔. If this is a ❓ Question, could you please share more details about what you're trying to achieve and the specific version of YOLO you're using? Also, check if you've upgraded to the latest version to ensure you're using the most up-to-date features and information:

pip install -U ultralytics

For reference, the augmentation details you mentioned might differ between YOLO versions. We suggest reviewing the detailed explanation under Training Settings in the docs. Feel free to let us know if anything is still unclear!

Join the Ultralytics community where it suits you best. For real-time chat, head to Discord 🎧. Prefer in-depth discussions? Check out Discourse. Or dive into threads on our Subreddit to discuss ideas and share insights with others.

Upgrade

Upgrade to the latest ultralytics package including all requirements in a Python>=3.8 environment with PyTorch>=1.8 to verify your issue is not already resolved in the latest version:

pip install -U ultralytics

Environments

YOLO may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all Ultralytics CI tests are currently passing. CI tests verify correct operation of all YOLO Modes and Tasks on macOS, Windows, and Ubuntu every 24 hours and on every commit.

⚠️ This is an automated response, but don't worry! An Ultralytics engineer will also review your discussion and assist you soon. 😊

[glenn-jocher]

@Zvyozdo4ka the augment parameter in YOLOv8 was a legacy boolean flag to enable basic augmentations, while auto_augment in YOLOv11 specifies advanced augmentation policies like 'randaugment' or 'autoaugment'. For YOLOv11, use auto_augment alongside task-specific params like hsv_h and flipud from our Train mode docs. The legacy augment flag is deprecated in YOLOv11.

[Zvyozdo4ka]

@glenn-jocher, @UltralyticsAssistant
thank you for your responses. I do not need any augmentation during the training, so I thought about shortening command and get rid of useless or default hyperparameters. So, i set like this, that's enough to disable the augmentation, right?

     auto_augment=None \
     close_mosaic=0 \
     hsv_h=0.0 \
     hsv_s=0.0 \
     hsv_v=0.0 \
     translate=0.0 \
     scale=0.0 \
     fliplr=0.0 \
     mosaic=0.0 \
     erasing=0.0 \

And also my image size is 500x300, and i wanna keep the original size, but yolo resizes them to 224 by default, or if i set imgsz=500, images are resized to 512x512. So I am wondering how to prevent augmentation. I also set shear=0.0, but as I can see it is also by default 0.0. I wanted to compare accuracy results, to understand which way is better for training.

[Y-T-G]

Training uses square imgsz. You can use rect=True to keep the rectangle size but it will use the nearest imgsz divisible by 32.

[glenn-jocher]

@Zvyozdo4ka the auto_augment parameter controls predefined augmentation policies (like RandAugment), while individual augment params (hsv, flip, etc) are manual settings. To fully disable all augmentations in YOLOv11, you need to set all these params:

degrees=0.0
perspective=0.0
flipud=0.0
mixup=0.0
copy_paste=0.0

along with those you already disabled. For image size, set imgsz=500 and add rect=True to maintain aspect ratio while training on rectangular images. See the augmentation docs for details: https://docs.ultralytics.com/modes/train/#augmentation-settings-and-hyperparameters

[Y-T-G]

augment is not applicable to training. It's for prediction.

auto_augment is applicable to classification models