Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

faster hsv augmentation #1110

Merged
merged 1 commit into from
Apr 28, 2020
Merged

Conversation

jveitchmichaelis
Copy link
Contributor

@jveitchmichaelis jveitchmichaelis commented Apr 28, 2020

Faster HSV augmentation, as per #1096

πŸ› οΈ PR Summary

Made with ❀️ by Ultralytics Actions

🌟 Summary

Enhanced color augmentation in image processing for better performance and stability.

πŸ“Š Key Changes

  • Introduced separate lookup tables (LUTs) for hue, saturation, and value channels in the augment_hsv function.
  • Changes include conversion from a single-step transformation to a multi-step process utilizing cv2.split and cv2.LUT for better control over image transformation.
  • Replaced direct application of random gains on the HSV image with an explicit calculation using lookup tables.

🎯 Purpose & Impact

  • 🎨 Improved Color Augmentation: By using lookup tables, the augmentation process becomes more stable and prevents artifacts caused due to direct multiplication.
  • πŸ› οΈ Precision Enhancement: More precise control over the HSV color space adjustments for hue, saturation, and value is now possible.
  • πŸƒβ€β™‚οΈ Potential Efficiency Gain: May lead to performance improvements in image pre-processing, enhancing the overall efficiency of machine learning workflows.
  • πŸ” Better Image Quality: Ensures high-quality image transformations, possibly leading to improved object detection by YOLOv3 algorithms.

@glenn-jocher
Copy link
Member

Awesome, thanks!

@glenn-jocher glenn-jocher merged commit 992d8af into ultralytics:master Apr 28, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants