How to remove a detection layer?

[nachoogriis]

Search before asking

• I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

I am using YOLOv5 to create a detection model. The model will only see small objects (boxes of 10x10). Taking this into account, I have thought of removing some of the detection layers that are in charge of detecting bigger objects in the image (in particular, I am thinking of keeping layers until P3/8). I have two questions:

1. The main reason I want to do this is to reduce the number of parameters and, therefore, making the model faster to train. Would this be beneficial? Would it help to detect better these small objects?

2. How could I implement this. I have looked at other issues such as Issue #1418 and try to implement the same backwards (removing the layers in charge of those bigger objects), but I face some errors which I am not sure how to fix. I have done the following changes.

Could someone please help me with this?

Additional

No response

[github-actions]

👋 Hello @nachoogriis, thank you for your interest in YOLOv5 🚀! 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>=3.8.0 with all requirements.txt installed including PyTorch>=1.8. To get started:

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

Environments

YOLOv5 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 YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 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]

@nachoogriis hello! Thanks for reaching out and diving deep into YOLOv5 customization for your project. 🚀

To address your questions:

1. On Model Parameter Reduction: Reducing the number of parameters by removing certain layers dedicated to detecting larger objects can indeed make the model faster to train and potentially more specialized on smaller objects. However, the benefit in detection accuracy for small objects isn't guaranteed and would largely depend on your specific dataset and experimental setup.

2. Implementing Layer Removal: Customizing the YOLOv5 architecture by removing particular layers (like larger object detection layers) requires careful adjustment of the configuration and potentially the model's forward method to ensure consistency throughout the network. Your approach to modify the network is on the right track. Here’s a simplified guideline:

   • In the model configuration file (*.yaml), remove or comment out the layers/parts you're interested in disabling.
   • Ensure that subsequent layers are correctly receiving the expected input sizes from the modified architecture. This often requires adjusting layer indices and connections accordingly.

   When facing errors, specifically dissect the error message to understand if it's due to mismatched dimensions or other configuration inconsistencies. Modifying deep learning models can be tricky, and often the devil is in the details.

Remember, making such modifications requires a strong understanding of the model's architecture and a good amount of trial and error. Keep experimenting, and use the error messages as clues for the next steps to take!

If persistent issues arise or if you have more complex customization needs, consider referring to the Ultralytics Docs for deeper insights or reaching out for more specialized support.

Wishing you success on your YOLOv5 journey! 🌟

[github-actions]

👋 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:

• Docs: https://docs.ultralytics.com
• HUB: https://hub.ultralytics.com
• Community: https://community.ultralytics.com

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 ⭐