The issue of antibiotic resistance poses a serious challenge in global health, prompting the search for safer and more effective alternative treatments. Garlic (Allium sativum) is known to have antimicrobial properties thanks to its allicin content, an active compound capable of inhibiting the growth of various microorganisms. This study aims to test the effectiveness of garlic as an antibiotic against oral microbes using an Object Detection Machine Learning technology approach with the viable plate count (VPC) method. The results show that higher concentrations of garlic significantly reduce the number of microbial colonies, with the highest effectiveness at a concentration of 50%. The machine learning model using object detection showed high precision (up to 100%) but low recall at a threshold of 20%, and more balanced performance at a threshold of 1% (precision 90.6%, recall 94.1%). These findings indicate that a machine learning-based VPC system can be a reliable and efficient alternative in microbiological analysis, and supports the potential of garlic as a sustainable natural antibiotic.