Tomato Yellow Leaf curl virus(TYLCV) is one of the most common disease in greenhouse tomato farming in South Korea. Detecting TYLCV symptom using only image is a challenge area for smart farm and inexpert farmer. In this study, we try to developed deep-learning classification model between healthy plant and TYLCV symptom. PSI system used for generating high-throughput Images, micro-tom samples were grown in phytotron which has environment condition control(LED light, CO2 Level, temperature and humidity). TYLCV sap inoculation of micro-tom processed on the 15th day after germination. Sample images were captured every 8 hours until day 25 from sap inoculation. we separated 2 phase by sample preprocess for machine learning. Only cropped by plant sample images of TYLCV and Healthy were imported Convolution neural network(CNN) model on first phase. Shattered and manually annotated images for healthy leaves and TYLCV symptom were imported CNN model on second phase. Result of first phase CNN model shows accuracy of 49.99% for classifying TYLCV infected plant. Second phase CNN model shows 99.95% accuracy for classifying TYLCV symptom leaves.