Detection mature bud for daylily based on Faster R-CNN integrated with CBAM

The daylily ( Hemerocallis citrina Baroni ) is rich in not only nutrition ingredients but also functional components, and the edible part flower, containing its pedicel. primary challenge developing a robotic harvester recognizing mature bud unstructured uncertain environment. objective of this study to propose an accurate detection model. citrina cv. ‘DatongHuanghua’ variety used study. We initially adopt VGG16, VGG19, ResNet50, ResNet101 ResNet152 as backbones Faster R-CNN respectively build different models. experimental results show that VGG19 ResNet50 are two best-performing models corresponding VGGNet ResNet, Average Precision (AP) 90.18%, while 88.35%. Based on these, we further integrate Convolutional Block Attention Module (CBAM) with three integration modes: plugging CBAM behind Conv5_x respectively, well between every “bottleneck” blocks ResNet50. comparison demonstrate best mode, model has 2.22% highest increase AP. Therefore, empirically validate performance for based integrated CBAM.