Attention Augmented Convolutional Networks

Benchmark Model Rank Results
image-classification-on-cifar-100AA-Wide-ResNet#143Percentage correct: 81.6
image-classification-on-imagenetAA-ResNet-152#812Top 1 Accuracy: 79.1%
object-detection-on-coco-test-devAA-ResNet-101 (RetinaNet)–box mAP: 39.2GFLOPs: 245