December 2023 Algorithms release
Last updated
Last updated
✨ We are thrilled to announce that in just 30 days, we will release an updated version of our Redness, Hydration, Uniformness, and Eye bags algorithms. Our team has been working diligently to retrain and improve these algorithms to provide you with faster and more accurate results.
✅These updates will be delivered automatically and require no additional actions from your side.
✨ The Quality algorithm gets an update as well! Read more below 🔽
⚠️ At the same time, we are will stop supporting sagging and wrinkles algorithms. We encourage you to use lines instead (read more about lines here).
What is changing? The update will focus on adding sub-metrics related to lighting intensity and direction.
What should I do?
✅If you use our B2C app, no actions are required as the changes will be released automatically to all users.
✅If you use our API and only use the main metrics, no actions are required.
⚠️If you use sub-metrics of the quality algorithm, please contact our support desk for assistance.
As previously announced a year ago, starting from the release of our updated algorithms, we will no longer be supporting wrinkles and sagging algorithms. Instead, we encourage you to use the lines algorithm, which is more accurate and contains both wrinkles and sagging information.
Redness
The update will focus on speeding up the analysis and improving the accuracy for some edge cases, resulting in changes of no more than ±5 points for majority cases.
Hydration
The update will focus on speeding-up the analysis and improving the accuracy for some edge cases, resulting in changes of no more than ±10 points for majority cases.
Uniformness
The update will focus on speeding-up the analysis and improving the accuracy, resulting in changes of no more than ±10 points for majority cases.
Eye bags
The update will focus on improving the accuracy and speed of the algorithm. We have also changed the logic of the overall eye condition score and improved the accuracy of eye bags and dark circles segmentation.