ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks

8 Oct 2019Qilong WangBanggu WuPengfei ZhuPeihua LiWangmeng ZuoQinghua Hu

Recently, channel attention mechanism has demonstrated to offer great potential in improving the performance of deep convolutional neural networks (CNNs). However, most existing methods dedicate to developing more sophisticated attention modules for achieving better performance, which inevitably increase model complexity... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT LEADERBOARD
Image Classification ImageNet ECA-Net (ResNet-152) Top 1 Accuracy 78.92% # 76
Top 5 Accuracy 94.55% # 51
Number of params 57.40M # 2
Image Classification ImageNet ECA-Net (ResNet-101) Top 1 Accuracy 78.65% # 80
Top 5 Accuracy 94.34% # 58
Number of params 42.49M # 2
Image Classification ImageNet ECA-Net (ResNet-50) Top 1 Accuracy 77.48% # 96
Top 5 Accuracy 93.68% # 67
Number of params 24.37M # 2
Image Classification ImageNet ECA-Net (MobileNetV2) Top 1 Accuracy 72.56% # 133
Top 5 Accuracy 90.81% # 95
Number of params 3.34M # 2