Enum QualityLevel

Namespace

AiDotNet.Enums

Assembly

AiDotNet.dll

Quality levels for adaptive inference on resource-constrained devices.

public enum QualityLevel

Fields

High = 2

High quality, maximum accuracy - prioritizes accuracy over speed. Uses minimal optimizations to preserve model quality.

Low = 0

Low quality, maximum speed - prioritizes performance over accuracy. Uses aggressive optimizations like quantization and layer skipping.

Medium = 1

Medium quality, balanced - good compromise between speed and accuracy. Uses moderate optimizations.

Remarks

For Beginners: When running AI models on devices with limited resources (like phones or edge devices), you often need to trade off quality for speed or battery life. This enum lets you choose how the model should balance these tradeoffs:

• Low: Fastest inference, uses least battery, but lower accuracy. Good for when battery is low or the device is under heavy load.
• Medium: Balanced - decent speed and accuracy. Good default for most situations.
• High: Best accuracy, slower inference, uses more battery. Use when you need the best results and the device has plenty of power.

The library can automatically switch between these levels based on battery level and CPU load.