基于改进MobileNetV2的棉花颜色分级检测OACSTPCD

In order to solve the problems that sensory inspection was easily influenced by subjective factors and unstable instrument inspection in cotton color grade detection,a method of using improved MobileNetV2 neural network to realize cotton color grade detection was proposed.Through the self-designed image acquisition device,5 kinds of cotton color samples from white cotton grade 1 to white cotton grade 5 were collected,and the data set was made.The features of the last three layers of MobileNetV2 network were fused and embedded into CBAM attention mechanism.Meanwhile,it was compared with GhostNet,ShuffleNetV2 and the original MobileNetV2 model to predict cotton color grade.The results showed that the improved MobileNetV2 accuracy in the test set was achieved 92.10%,which was 3.01 percentage points higher than GhostNet,4.61 percentage points higher than ShuffleNetV2,and 1.24 percentage points higher than the original mobileNetV2.It has better detection effect.