At present, the process of building AI (artificial intelligence) is as follows: data collection → data cleaning → data annotation → data storage → feature analysis → intelligent application. Among these steps, data collection and analysis are the main focuses. The following are examples of AI application!
Smart Agriculture– Tomato Harvesting
In the vast tomato garden, tomatoes come in different colors and sizes. Xianjun Zheng and his team, the BEEHIVE Data Technology, use OpenCV and YOLO technology to measure and label tomatoes. Data collected can be used as the baseline for the visual sensor in the autonomous harvesting system. After testing, data collecting, and adjusting the program, the selection process has become more efficient, and the autonomous vehicle is able to pick out ripe fruits by itself.
Smart Healthcare- Robots Prepare Your Medicines
According to the Ministry of Health and Welfare, every elderly in Taiwan takes an average of 5 to 6 drugs per day. To promote better drug use, big data analysis is used in patient diagnosis, prescriptions, consultations, taking drugs, etc. In a field test in a pharmacy, Zenbo, a robot developed by the team from Tajen University, can accurately prepare the cold medicine based on the patient’s description of symptoms. The robot can also warn patients against mixing drugs to avoid adverse effects on patients due to drug interactions.
In addition, professor Che Wei Lin and his team from the Department of Biomedical Engineering, National Cheng Kung University, have also developed an AI automatic pill-dispenser named UTOPIL, named after Utopia, the imaginary, ideal civilization. The dispensing mechanism consists of image recognition and a vacuum pump. A database can, therefore, be built with the results of pill identification. According to the needs of the elderly, the dispenser can help patients take the correct medications on time.
Smart Finance– Smarter Investments
How to make the most profit through a large amount of trading information in the ever-changing financial market? A team led by Professor Xinjia Huang from National Kaohsiung University of Science and Technology has the answer. The team uses the fund as an asset and builds a “Smart Financial Management System” by establishing a net asset value database through the quantitative indicators on anuefund.com. This system adopts robotic deep learning to analyze the past financial data and indicators provided by financial experts so that the system can provide portfolio structures for three types of investors – aggressive, moderate and conservative, and recommend investors the best timing to buy and sell.
The system is currently in the testing stage, but it has reached 90% winning percentage! More importantly, using data in the past, the system can conduct backtesting for investors to keep track of the performance of their investments!