How can I see F1-score and precision/recall curves for validation set ?

Grhanas commented 1 month ago

Search before asking

[X] I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

Hello! I can see the F1-score for the training dataset after completing the training process. Additionally, I have access to the dfl_loss, box_loss, and cls_loss for both the training and validation sets, as shown in "results.png." However, I'm unable to find precision and recall values for the validation set. I would like to understand if my model is experiencing overfitting or underfitting.

Additional

No response

UltralyticsAssistant commented 1 month ago

👋 Hello @Grhanas, thank you for reaching out with your question! 🚀

To get started with understanding your validation metrics, you might want to check out our ⭐️ Tutorials where we cover various aspects including Custom Data Training and more.

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

For training-related questions, ensure you're following our Tips for Best Training Results. Including more details like dataset examples and full training logs can be very helpful.

Requirements

Ensure you are using Python>=3.8.0 with all dependencies from requirements.txt, including PyTorch>=1.8. Here's a quick setup guide:

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

Environments

YOLOv5 can be run in various environments. Some options include:

Notebooks with free GPU:
Google Cloud Deep Learning VM: Learn more in the GCP Quickstart Guide
Amazon Deep Learning AMI: Refer to the AWS Quickstart Guide
Docker Image: Follow the Docker Quickstart Guide

Status

Current CI status: If this badge is green, all tests are passing.

Discover YOLOv8 🚀

Explore the capabilities of our latest model YOLOv8 for even better performance in object detection, image segmentation, and classification. Check the YOLOv8 Docs and get started with this command:

pip install ultralytics

This is an automated response, but an Ultralytics engineer will review and assist you soon. Thank you for your patience! 😊

pderrenger commented 1 week ago

@Grhanas to view precision, recall, and F1-score curves for the validation set, you can check the results.png file generated in the runs/train/exp directory after training. This file includes precision-recall (PR) curves, which can help you assess model performance and overfitting.

ultralytics / yolov5