【ML】The List of Image Classification Models

List some CNN models that are suitable for image classification and these numbers of parameters.

ResNet

ResNet-18: ~11.7 million parameters
ResNet-34: ~21.8 million parameters
ResNet-50: ~25.6 million parameters
ResNet-101: ~44.5 million parameters
ResNet-152: ~60 million parameters

VGG

VGG-16: ~138 million parameters
VGG-19: ~143 million parameters

Inception Networks

Inception v1 (GoogleNet): ~6.8 million parameters
Inception v3: ~23.8 million parameters

DenseNet

DenseNet-121: ~8 million parameters
DenseNet-169: ~14 million parameters
DenseNet-201: ~20 million parameters
DenseNet-161: ~28.7 million parameters
DenseNet-264: ~33 million parameters

MobileNets (parameters can vary significantly with changes in alpha and resolution multipliers)

MobileNetV1 (1.0 224): ~4.2 million parameters
MobileNetV2 (1.0 224): ~3.5 million parameters
MobileNetV3 Large: ~5.4 million parameters

Vision Transformers (ViT)

ViT-B/16 (base model with patch size 16x16): ~86 million parameters
maxvit_pico: 7.54M
maxvit_nano: 15.5M
maxvit_tiny: 33M
maxvit_base: 120M

Xception

Xception: ~22.9 million parameters

EfficientNet

EfficientNet-B0: ~5.3 million parameters
EfficientNet-B1: ~7.8 million parameters
EfficientNet-B2: ~9.2 million parameters
EfficientNet-B3: ~12 million parameters
EfficientNet-B4: ~19 million parameters
EfficientNet-B5: ~30 million parameters
EfficientNet-B6: ~43 million parameters
EfficientNet-B7: ~66 million parameters

Eva

eva02_tiny: 5.5M
eva02_small: 22.1M
eva02_base: 85.8M
eva02_large: 326M