hyperparameter evolution on a small dataset

[Lumcoin]

Search before asking

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

Question

I am using YOLOv5s for my thesis. One of the challenges in my thesis is using a very small train set consisting of 119 images and roughly 1500 objects with 5 different classes. I tried hyperparameter evolution using different epoch settings and the "best" result i got was using the recommended setting of 10 epochs. I unfortunately did not document the settings fully and thus I am uncertain how large the batch size was. In the worst case it might have been 1.

I execute a 16 x 16 grid search to find the optimal conf thres and iou thres combination. Then I use those thresholds to evaluate the models. Using bootstrapping the F1 score of the model trained with tuned hyperparameters performs worse than the stock hyperparameters for finetuning with a certainty of 99.995 % a.k.a. it performed worse.

Did I likely do something wrong or is this typical with such a small dataset? I am aware that at least 1500 images per class and 10000 instances per class are recommended for YOLOv5.

P.S. Is there a way of calculating a confidence interval of the F1 score or bootstrapping the F1 score distribution natively using YOLOv5? This should be best practice imho because a point estimate from a sample (unfortunately most common in literature) has very little meaning for the population. A point estimate also can't be used for hypothesis tests.

Additional

No response

[github-actions]

👋 Hello @Lumcoin, 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 screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://ultralytics.com or email support@ultralytics.com.

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

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle 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 (train.py), validation (val.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[glenn-jocher]

@Lumcoin hyperparameter evolution optimizes hyps to a specific base scenario (that you specify) and nothing else. Using the optimized hyps for a different scenario is on you.

See hyp evolution tutorial for details:

YOLOv5 Tutorials

• Train Custom Data  🚀 RECOMMENDED
• Tips for Best Training Results  ☘️ RECOMMENDED
• Weights & Biases Logging  🌟 NEW
• Supervisely Ecosystem  🌟 NEW
• Multi-GPU Training
• PyTorch Hub  ⭐ NEW
• TFLite, ONNX, CoreML, TensorRT Export 🚀
• Test-Time Augmentation (TTA)
• Model Ensembling
• Model Pruning/Sparsity
• Hyperparameter Evolution
• Transfer Learning with Frozen Layers  ⭐ NEW
• TensorRT Deployment

Good luck 🍀 and let us know if you have any other questions!

[myasser63]

@Lumcoin I am doing similar project for my thesis and even smaller dataset. Can you tell me the way you did grid search or you used the hyperparameter evolution by yolov5

[MattGrouchy]

@Lumcoin

Has there been any developments in your findings with hyperparameter evolution?

[github-actions]

👋 Hello, this issue has been automatically marked as stale because it has not had recent activity. Please note it will be closed if no further activity occurs.

Access additional YOLOv5 🚀 resources:

• Wiki – https://github.com/ultralytics/yolov5/wiki
• Tutorials – https://docs.ultralytics.com/yolov5
• Docs – https://docs.ultralytics.com

Access additional Ultralytics ⚡ resources:

• Ultralytics HUB – https://ultralytics.com/hub
• Vision API – https://ultralytics.com/yolov5
• About Us – https://ultralytics.com/about
• Join Our Team – https://ultralytics.com/work
• Contact Us – https://ultralytics.com/contact

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 YOLOv5 🚀 and Vision AI ⭐!