VGG-16

Click a block to read its description, or select it with Tab and Enter.

VGG-16Classification, 224 × 224 input, 1,000 classes. Tensor sizes exclude batch.LibreYOLOVGG-16Classification, 224 × 224 input, 1,000 classes. Tensor sizes exclude batch.Stage 1Input3 × 224 × 224Conv2d 3×3 / 164 × 224 × 224; p=1, g=1ReLU64 × 224 × 224Conv2d 3×3 / 164 × 224 × 224; p=1, g=1ReLU64 × 224 × 224MaxPool2d64 × 112 × 112Stage 2Input64 × 112 × 112Conv2d 3×3 / 1128 × 112 × 112; p=1, g=1ReLU128 × 112 × 112Conv2d 3×3 / 1128 × 112 × 112; p=1, g=1ReLU128 × 112 × 112MaxPool2d128 × 56 × 56Stage 3Input128 × 56 × 56Conv2d 3×3 / 1256 × 56 × 56; p=1, g=1ReLU256 × 56 × 56Conv2d 3×3 / 1256 × 56 × 56; p=1, g=1ReLU256 × 56 × 56Conv2d 3×3 / 1256 × 56 × 56; p=1, g=1ReLU256 × 56 × 56MaxPool2d256 × 28 × 28Stage 4Input256 × 28 × 28Conv2d 3×3 / 1512 × 28 × 28; p=1, g=1ReLU512 × 28 × 28Conv2d 3×3 / 1512 × 28 × 28; p=1, g=1ReLU512 × 28 × 28Conv2d 3×3 / 1512 × 28 × 28; p=1, g=1ReLU512 × 28 × 28MaxPool2d512 × 14 × 14Stage 5Input512 × 14 × 14Conv2d 3×3 / 1512 × 14 × 14; p=1, g=1ReLU512 × 14 × 14Conv2d 3×3 / 1512 × 14 × 14; p=1, g=1ReLU512 × 14 × 14Conv2d 3×3 / 1512 × 14 × 14; p=1, g=1ReLU512 × 14 × 14MaxPool2d512 × 7 × 7ClassifierAdaptiveAvgPool2d512 × 7 × 7Flatten25,088Linear25088 to 4096ReLU4096Dropout (eval identity)p=0.5; 4096Linear4096 to 4096ReLU4096Dropout (eval identity)p=0.5; 4096Linear4096 to 1000Continue at stage 2Continue at stage 3Continue at stage 4Continue at stage 5Continue at classifierConv2d: bias=True, 3×3, stride 1, padding 1. MaxPool2d: 2×2, stride 2, padding 0.Stage inputs are the preceding stage outputs. Dropout is identity in eval. Output: 1,000 logits.Source: libreyolo/models/vgg/nn.py. Revision a4d0ecc9e17f.libreyolo.com