The table prvoides the models and results of various models on CIFAR100. Learning rate =0.1 and will be divided by 10 every 70 epochs. Total 300 epochs. Using SGD optimizer, momentum=0.9, weight_decay=5e-4. Loss is CrossEntropyLoss. Batch-size=512.
This repository is a PyTorch implementation of our coordinate attention (will appear in CVPR2021).
Code for the paper MSAF: Multimodal Split Attention Fusion. This is our implementation of the MSAF module and the three MSAF-powered multimodal networks.
Figure 1. Illustration of our DCANet. We visualize intermediate feature activation using Grad-CAM. Vanilla SE-ResNet50 varies its focus dramatically at different stages. In contrast, our DCA enhanced SE-ResNet50 progressively and recursively adjusts focus, and closely pays attention to the…