[17a-A37-1~8] CS.2 Code-sharing Session of 3.2 & 4.4

9:45 AM - 10:00 AM JST(12:45 AM - 1:00 AM UTC)English Presentation

[17a-A37-3] Enhancing the Accuracy of Identification in Complex Environmental Backgrounds using YOLO V7 and U2NET: Orchid Repotting

〇(M2)HUNG WEI HSU¹, Chih-Chung Wang¹, Jia-Han Li¹ (1.National Taiwan University)

Keywords:

deep learning,Complex Environmental Backgrounds,YOLO v7

In the horticultural industry, determining the appropriate timing for repotting is crucial for plant health and growth. Traditional Automated Optical Inspection (AOI) techniques fall short in addressing the high variability inherent in horticulture due to environmental, genetic, and growth-related factors. These traditional methods are inadequate for dealing with the dynamic and complex nature of plant cultivation. Therefore, advanced deep learning models are needed to predict optimal repotting times and ensure precise target object detection amidst complex backgrounds. This study employs YOLOv7 for prediction and U2NET for image segmentation, replacing the masking function of YOLOv7. The effectiveness of three YOLOv7-trained models is compared to evaluate their performance in recognizing Oncidium orchids under challenging, cluttered background conditions.