Detection and recognition of tea buds by integrating deep learning and image-processing algorithm

The accurate detection of tea buds is a crucial foundation for achieving intelligent plucking of tea. However, in unstructured environments, the detection of these minuscule buds with extreme length-to-width ratios poses a significant challenge. In this study, a method was developed for the detection of tea buds in complex environments. At first, the YOLOv5s_DCV model was developed based on the YOLOv5s network model, which incorporates advanced techniques such as Deformable ConvNets V2, Content-Aware ReAssembly of Features, and Varifocal Loss, considering both efficiency and accuracy. Besides, we used image processing methods to reduce the model's sensitivity to changes in lighting conditions. The experimental results demonstrated that our method achieves impressive precision with an average precision (AP) of 90.6%, surpassing mainstream object detection methods. This study holds paramount theoretical and practical significance, offering robust support for the accurate detection and precise localization of tea buds, as well as phenotype identification and the accurate estimation of tea leaf yield.