Identification of Harvestable Tomatoes using Yolov8

The implementation of tomato picking robots holds immense importance for greenhouse tomato cultivation, as it significantly reduces labor requirements and enhances production efficiency. This study explores the application of Yolov8, a state-of-the-art object detection algorithm, for the precise identification of harvestable tomatoes in greenhouse settings. The yolov8 model is trained and fine-tuned using a comprehensive dataset comprising diverse tomato varieties that are ripe and unripe. The proposed methodology harnesses the power of deep learning to enable real-time detection of ripe tomatoes, facilitating timely and efficient harvesting practices. Leveraging the yolov8 architecture, the system demonstrates exceptional accuracy and speed, making it suitable for high-throughput greenhouse operations. Various challenges, including occlusion and varying lighting conditions, are addressed through robust training strategies, ensuring the model’s reliability in practical farming scenarios. The most advanced visual tomato identification technology focuses mostly on ripe tomatoes because they stand out from the backdrop in terms of color. Furthermore, the study evaluates the performance of the model against traditional computer vision methods, highlighting the superior accuracy and efficiency of the deep learning approach. The predictive accuracy of ripe and unripe tomatoes was found to be 97.23% when segmentation is applied in Yolov8. The integration of this advanced technology into greenhouse practices promises to revolutionize the way harvestable produce is identified and harvested, paving way for sustainable and efficient agricultural processes in future.