Detection of maize leaf diseases using improved MobileNet V3-small

In order to realize the intelligent identification of maize leaf diseases for accurate prevention and control, this study proposed a disease detection method based on improved MobileNet V3-small, using UAV collect images establish dataset in complex context, explored effects data expansion migration learning model recognition accuracy, recall rate, F1-score instructive evaluative indexes, results show that two approaches effectively accuracy model. The structured compression V3-small bneck layer retains only 6 layers, multiplier each was redesigned, 32-fold fast downsampling used first layer, location SE module optimized. had an average 79.52% test set, 77.91%, 78.62%, size 2.36 MB, single image speed 9.02 ms. can meet requirements mobile or embedded devices. This provides technical support realizing diseases. Keywords: disease, recognition, compression, MobileNetV3-small DOI: 10.25165/j.ijabe.20231603.7799 Citation: Gao A, Geng A J, Song Y P, Ren L L, Zhang Y, Han X. Detection V3-small. Int J Agric & Biol Eng, 2023; 16(3): 225–232.