Dynamic Convolution: Attention over Convolution Kernels

Open paper
Benchmark Model Rank Results
image-classification-on-imagenetDY-MobileNetV2 ×1.0#1019Top 1 Accuracy: 74.4%Number of params: 11.1MGFLOPs: 0,626
image-classification-on-imagenetDY-MobileNetV2 ×0.75#1039Top 1 Accuracy: 72.8%Number of params: 7MGFLOPs: 0.435
image-classification-on-imagenetDY-ResNet-18#1041Top 1 Accuracy: 72.7%Number of params: 42.7MGFLOPs: 3.7
image-classification-on-imagenetDY-MobileNetV3-Small#1077Top 1 Accuracy: 69.7%Number of params: 4.8MGFLOPs: 0.137
image-classification-on-imagenetDY-MobileNetV2 ×0.5#1080Top 1 Accuracy: 69.4%Number of params: 4MGFLOPs: 0.203
image-classification-on-imagenetDY-ResNet-10#1090Top 1 Accuracy: 67.7%Number of params: 18.6MGFLOPs: 1.82
image-classification-on-imagenetDY-MobileNetV2 ×0.35#1099Top 1 Accuracy: 64.9%Number of params: 2.8MGFLOPs: 0.124

Rank counts only results with a code link.