The Procyon AI benchmark measures inference times on popular models like MobileNet, ResNet, and YOLO. It's designed to show how well a system can accelerate machine learning tasks in real use.
In UL Procyon AI Computer Vision, the H05 scores 125 overall, and the per-test times include 1.34 (MobileNet V3), 10.11 (ResNet 50), 32.17 (Inception V4), 35.48 (DeepLab V3), 73.67 (YOLO V3), and 3557.55 (Real-ESRGAN). The lighter classification workloads finish quickly, while Real-ESRGAN is the clear outlier and will, as expected, take a long time on integrated graphics.