Turmeric is a valuable crop, but it is prone to various diseases that can significantly impact its production. Early detection of these diseases is crucial to prevent crop failure and losses. In this research, we propose a novel approach for accurately identifying turmeric plant diseases using a single-phase detection model based on machine learning. Our method, called the Improved YOLOV3-Tiny model, leverages a residual network structure and convolutional neural networks to improve detection accuracy compared to the traditional YOLOV3-Tiny model. We tested our method on images captured during both day and night and found that it outperformed other methods such as YOLO and Quicker R-CNN with the VGG16 prototypical. We also found that augmenting the turmeric leaf dataset with Cycle-GAN improved detection accuracy, particularly for smaller datasets. Overall, our proposed model offers both high accuracy and fast recognition speed, making it a valuable tool for detecting turmeric leaf diseases.