JOYY has patented a method for training a face detection model using three-dimensional reconstruction and UV coordinate mapping. The process involves extracting, inputting, and calculating to optimize the semantic segmentation network. The innovative approach aims to enhance face detection accuracy and efficiency. GlobalData’s report on JOYY gives a 360-degree view of the company including its patenting strategy. Buy the report here.

According to GlobalData’s company profile on JOYY, was a key innovation area identified from patents. JOYY's grant share as of February 2024 was 47%. Grant share is based on the ratio of number of grants to total number of patents.

Training face detection model using three-dimensional reconstruction and uv coordinates

Source: United States Patent and Trademark Office (USPTO). Credit: JOYY Inc

A recently granted patent (Publication Number: US11922707B2) discloses a method for training a face detection model that involves acquiring a training face image, performing three-dimensional reconstruction based on a preset model, generating a training UV coordinate map, and inputting the image into a semantic segmentation network. The method includes calculating a loss rate, judging gradient conditions, and performing gradient descent on network parameters. The patent also covers a method for detecting face key points using a pre-trained face detection model and a UV template image. Additionally, the patent describes a device and a computer-readable storage medium for executing the training and detection methods.

Furthermore, the patent details the process of three-dimensional reconstruction using principal component analysis and eigenvalue matrices. It also includes methods for aligning and preprocessing three-dimensional face models before analysis. The patent further discloses a method for generating a training UV coordinate map based on three-dimensional face models and a spherical surface projection technique. The patent also covers the iterative optimization process for eigenvalue matrices and projection parameters using a stochastic gradient descent method. Overall, the patent provides a comprehensive approach to training face detection models and detecting face key points using advanced techniques and algorithms, aiming to enhance accuracy and efficiency in facial recognition technology.

To know more about GlobalData’s detailed insights on JOYY, buy the report here.

Premium Insights


The gold standard of business intelligence.

Blending expert knowledge with cutting-edge technology, GlobalData’s unrivalled proprietary data will enable you to decode what’s happening in your market. You can make better informed decisions and gain a future-proof advantage over your competitors.


GlobalData, the leading provider of industry intelligence, provided the underlying data, research, and analysis used to produce this article.

GlobalData Patent Analytics tracks bibliographic data, legal events data, point in time patent ownerships, and backward and forward citations from global patenting offices. Textual analysis and official patent classifications are used to group patents into key thematic areas and link them to specific companies across the world’s largest industries.