US 12,087,046 B2
Method for fine-grained detection of driver distraction based on unsupervised learning
Jie Chen, Hefei (CN); Bing Li, Hefei (CN); Zihan Cheng, Hefei (CN); Haitao Wang, Hefei (CN); Jingmin Xi, Hefei (CN); and Yingjian Deng, Hefei (CN)
Assigned to Anhui University, Hefei (CN)
Filed by Anhui University, Hefei (CN)
Filed on Apr. 28, 2022, as Appl. No. 17/661,177.
Claims priority of application No. 202111527027.2 (CN), filed on Dec. 14, 2021.
Prior Publication US 2023/0186436 A1, Jun. 15, 2023
Int. Cl. G06V 10/82 (2022.01); G06V 20/59 (2022.01); G06V 20/70 (2022.01)
CPC G06V 10/82 (2022.01) [G06V 20/597 (2022.01); G06V 20/70 (2022.01)] 10 Claims
OG exemplary drawing
 
1. A method for fine-grained detection of driver distraction based on unsupervised learning, comprising the following steps:
acquiring distracted driving image data; and
inputting the acquired distracted driving image data into an unsupervised learning detection model, analyzing the distracted driving image data by using the unsupervised learning detection model, and determining a driver distraction state according to an analysis result, wherein
the unsupervised learning detection model comprises a backbone network, projection heads, and a loss function;
the backbone network is a RepMLP network structure incorporating a multilayer perceptron (MLP);
the projection heads are each an MLP incorporating a residual structure; and
the loss function is a loss function based on contrastive learning and a stop-gradient strategy.