This page summarizes the observed frame rates for the benchmarked models. All values are frames per second (FPS).

Results

Model	Task	MIPI image	SPI: JPEG input tensor + output tensor	SPI: output tensor only
YOLOv8n	Object detection	30	3.5	15
YOLOv8n-pose	Pose estimation	30	3.5	10
YOLOv8n-cls	Image classification	30	2.13	30
YOLOv8n-seg	Instance segmentation	30	3.3	7.5
SSD MobileNet	Object detection	30	3.3	15
HigherHRNet	Pose estimation	30	2.45	4
MobileNetV2	Image classification	30	6.5	30
DeepLabV3+	Semantic segmentation	30	2	4.46

Metric Definitions

MIPI image is the image frame rate delivered through the MIPI image path.
SPI: JPEG input tensor + output tensor is the frame rate when SPI metadata includes both the JPEG input tensor and the model output tensor.
SPI: output tensor only is the frame rate when SPI metadata includes only the model output tensor.

MIPI image FPS and SPI tensor FPS describe different data paths. Use the MIPI result to evaluate image delivery and the SPI result that matches the metadata mode used by your application.

Reproducing SPI Measurements

The Pico 2 inference FPS benchmark measures inference-frame delivery without parsing, forwarding, or retaining the metadata payload. It supports both SPI modes shown above and documents its build options, defaults, and reported metrics.