Intelligent Medicinal Plant Factory

The development of medicinal plants has been annually increasing with a projected market of 1.11 billion USD in the future. The majority of medicinal plants in Taiwan relies on imports, which can be mixed in quality and contain various contaminants, such as pesticides, heavy metals and other pollutants, or sales of incorrect substitutes, as well as the problem of inefficacy or taking the wrong medicine. In order to avoid these problems, this research develops a set of methods for following the production history of hydroponic or soil cultivated plants and quality inspection throughout the entire process of planting to final delivery, as Good Agricultural and Collection Practices (GACP) was used as the resume architecture of medicinal plants, while the Internet of Things (IoT) under networking technology would help to collect and control data. By experiments, the optimal growth parameters of medicinal plants were found and could be used to carry out preventive automatic control by C4.5 algorithm. During planting and after harvest, image classification and detection were achieved by faster R-CNN and Inception Resnet V2, as the plant's growth cycle was compared to confirm if medicinal plants were consistent with their optimal growth conditions, as well as predicting the nutrient content. The production history and IoT system experiment have established in the cooperation factory for one month which can be remotely monitored. The cultivated plants are mainly Taiwan Jewel Orchid, with about 10,000 plants. With artificial intelligence (AI), the system is possible to identify clear leaves.