Multiscale Feature Filtering Network for Image Recognition System in Unmanned Aerial Vehicle

<div>Top-1 testing error rate versus number of epochs for CIFAR-100. The numbers of parameters for ResNet-101 and MFFNeXt-29 (2c × 64) are 42.70 M and 9.37 M, respectively.</div>

Complexity

fig7

Figure 7

Figure 7: Multiscale Feature Filtering Network for Image Recognition System in Unmanned Aerial Vehicle 