RevBiFPN: The Fully Reversible Bidirectional Feature Pyramid Network

Open paper
Benchmark Model Rank Results
image-classification-on-imagenetRevBiFPN-S6#330Top 1 Accuracy: 84.2%Number of params: 142.3MGFLOPs: 38.1
image-classification-on-imagenetRevBiFPN-S5#386Top 1 Accuracy: 83.7%Number of params: 82MGFLOPs: 21.8
image-classification-on-imagenetRevBiFPN-S4#464Top 1 Accuracy: 83%Number of params: 48.7MGFLOPs: 10.6
image-classification-on-imagenetRevBiFPN-S3#637Top 1 Accuracy: 81.1%Number of params: 19.6MGFLOPs: 3.33
image-classification-on-imagenetRevBiFPN-S2#752Top 1 Accuracy: 79%Number of params: 10.6MGFLOPs: 1.37
image-classification-on-imagenetRevBiFPN-S1#872Top 1 Accuracy: 75.9%Number of params: 5.11MGFLOPs: 0.62
image-classification-on-imagenetRevBiFPN-S0#930Top 1 Accuracy: 72.8%Number of params: 3.42MGFLOPs: 0.31

Rank counts only results with a code link.