Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Abstract. Tea (Camellia sinensis) is a high-value cash crop that produces a huge market value. Suitable temperature and relative humidity are critical factors to tea tree growing. Furthermore, in some unfavorable weather conditions, disease outbreaks might occur. With lesions arising, adverse impacts cause withering of tea leaves and results in the reduction in yield and profit. Thereby, early detection or on-site monitoring can provide effective integrated pest management (IPM) strategies to control the infected area and prevent further yield decreasing. In recent years, object detection using traditional image processing has been gradually replaced by convolutional neural network (CNN) due to its capability to identify targets with high complexity with a faster calculation speed. In this study, more than 1000 images of tea leaves are used to train the model based on faster region-based convolutional neural network (Faster R-CNN). The proposed model classifies three types of tea diseases, including brown blight, blister blight, and algal leaf spot. Preliminary results with 223 testing images performs an average precision (AP) of 63.58%, 81.08%, 64.71% for the identification of brown blight, blister blight, and algal leaf spot, respectively. The proposed algorithm provides tea farmers a convenient tool to identify the occurrence of three tea diseases in field automatically.