An Online Semisupervised Learning Model for Pedestrians’ Crossing Intention Recognition of Connected Autonomous Vehicle Based on Mobile Edge Computing Applications

Table 3

Accuracy of pedestrian crossing intention recognition based on SVM, S3VM, and OS3VM.

Labeled sample proportion	SVM	S^3VM	OS^3VM
10	82.43	85.93	89.16
20	84.57	87.42	91.38
30	86.12	88.98	92.47
40	87.86	90.50	94.83
50	88.32	91.07	94.66
60	88.45	91.21	94.72
70	89.27	91.43	94.65