A Research on Automatic Identification of Early Crop Diseases and Pests Based on Attention Mechanism
In response to the problem of poor automatic recognition of early crop diseases and pests,the ShuffleNetV2 network is improved by introducing adaptive attention mechanisms at different positions in the network,and replacing the Relu activation func-tion in the original network with a Hardswish activation function.The ShuffleNetV2 ACECA network is designed to solve the prob-lem of automatic recognition of early crop diseases and pests.Experiments show that while maintaining the performance of the model,the Hardswish activation function can effectively reduce the number of parameters in the model.At the same time,the intro-duction of adaptive attention mechanism improves the recognition effect of the network on early pests and diseases,which has a more positive impact on production practice.
attention mechanismidentification of early crop diseases and pestsShuffleNetV2 network
NETL
NSTL
万方数据
