Transformer-Based Data-Driven Video Coding Acceleration for Industrial Applications

Table 1

Splitting accuracy of four transformer models at each CU depth.

Depth level	QP 22 (%)	QP 27 (%)	QP 32 (%)	QP 37 (%)
Depth 0	95.58	91.08	91.04	89.38
Depth 1	78.12	81.81	82.44	82.33
Depth 2	74.54	80.60	82.55	84.87