Enum EfficientNetVariant

Namespace

AiDotNet.Enums

Assembly

AiDotNet.dll

Specifies the EfficientNet model variant.

public enum EfficientNetVariant

Fields

B0 = 0

EfficientNet-B0: Base model (5.3M parameters, 224x224 input).

The baseline model, offering good accuracy with minimal compute.

B1 = 1

EfficientNet-B1: Scaled model (7.8M parameters, 240x240 input).

B2 = 2

EfficientNet-B2: Scaled model (9.2M parameters, 260x260 input).

B3 = 3

EfficientNet-B3: Scaled model (12M parameters, 300x300 input).

B4 = 4

EfficientNet-B4: Scaled model (19M parameters, 380x380 input).

B5 = 5

EfficientNet-B5: Scaled model (30M parameters, 456x456 input).

B6 = 6

EfficientNet-B6: Scaled model (43M parameters, 528x528 input).

B7 = 7

EfficientNet-B7: Largest model (66M parameters, 600x600 input).

The largest standard variant, offering state-of-the-art accuracy.

Custom = 8

Custom EfficientNet variant for testing with minimal layers.

Use this variant for unit tests to minimize construction time. Typically uses 32x32 input with 1.0 width/depth multipliers.

Remarks

EfficientNet variants use compound scaling to balance network depth, width, and resolution. Each variant (B0-B7) represents a different scale factor, with larger variants offering better accuracy at the cost of more computation.

For Beginners: The variant number (B0-B7) indicates the scale of the network. B0 is the smallest and fastest, while B7 is the largest with the highest accuracy. Choose based on your accuracy requirements and computational budget.