ultralytics/yolov3

Batch size in loss function

AlexeyDate opened this issue · 3 comments

AlexeyDate commented

Search before asking

Question

Thank you for such a useful repo!
I try to understand some features and I am thinking about calculation loss function. Why are you multiplying the calculated values by the batch size in this line: return (lbox + lobj + lcls) * bs, torch.cat((lbox, lobj, lcls)).detach()
I think you need to divide all values per batches to get the averages, no?

Additional

No response

github-actions commented

👋 Hello @AlexeyDate, thank you for your interest in YOLOv3 🚀! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide a minimum reproducible example to help us debug it.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset image examples and training logs, and verify you are following our Tips for Best Training Results.

Requirements

Python>=3.7.0 with all requirements.txt installed including PyTorch>=1.7. To get started:

git clone https://github.com/ultralytics/yolov5  # clone
cd yolov5
pip install -r requirements.txt  # install

Environments

YOLOv3 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

Status

YOLOv3 CI

If this badge is green, all YOLOv3 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv3 training, validation, inference, export and benchmarks on macOS, Windows, and Ubuntu every 24 hours and on every commit.

Introducing YOLOv8 🚀

We're excited to announce the launch of our latest state-of-the-art (SOTA) object detection model for 2023 - YOLOv8 🚀!

Designed to be fast, accurate, and easy to use, YOLOv8 is an ideal choice for a wide range of object detection, image segmentation and image classification tasks. With YOLOv8, you'll be able to quickly and accurately detect objects in real-time, streamline your workflows, and achieve new levels of accuracy in your projects.

Check out our YOLOv8 Docs for details and get started with:

pip install ultralytics
glenn-jocher commented

@AlexeyDate thank you for your kind words and for reaching out with your question! I appreciate your interest in understanding the loss function calculation in the YOLOv3 repository.

The multiplication of the loss values by the batch size in the mentioned line (return (lbox + lobj + lcls) * bs, torch.cat((lbox, lobj, lcls)).detach()) does not aim to calculate averages.

In object detection tasks, the loss function is typically computed per image. However, during training, it is common to process multiple images in a batch simultaneously. To properly scale the loss contribution from the batch, the loss values are multiplied by the batch size.

This scaling is done to ensure that gradients and updates to the model's parameters are appropriately adjusted when considering a batch of samples instead of a single sample. Dividing the loss values by the batch size would underestimate the impact of the loss on the overall model update.

I hope this clarifies the purpose of multiplying the loss values by the batch size. If you have any further questions or need additional assistance, please feel free to ask.

github-actions commented

👋 Hello there! We wanted to give you a friendly reminder that this issue has not had any recent activity and may be closed soon, but don't worry - you can always reopen it if needed. If you still have any questions or concerns, please feel free to let us know how we can help.

For additional resources and information, please see the links below:

Feel free to inform us of any other issues you discover or feature requests that come to mind in the future. Pull Requests (PRs) are also always welcomed!

Thank you for your contributions to YOLO 🚀 and Vision AI ⭐