"Drones offer a wider viewgreater mobilityflexibility. When combined with 5G sensors, a smart coastal patrolenvironmental sustainability system can be built. Its key problem is how to analyze both extremely smalllarge objects in aerial images at the same time. Features:
 
 1. Direct real-time analysis of aerial imagesenvironmental inspections.
 
 2. Actual systems have been well developed, tested,evaluated such as marine debris detection, oil pollution detection, coast guard system,so on.
 
 3. Large-scale investigations on endangered species such as white dolphinsenvironmental pollution in Taiwan,contributions to Taiwan＇s sustainable environment can be conducted."

"The key problem to build a drone-based inspection application is how to analyze both extremely smalllarge objects at the same time. We propose the concept of a parallel feature pyramid to achieve fasthigh accuracy detection of extremely small objects. It is published in the leading journal IEEE Trans. on IP. Features include: 
 
 1. Real-time direct analysis of aerial imagesenvironmental inspections.
 
 2. Actual systems such as marine debris detection, oil pollution detection, coast guard system,so on have been well developed, tested,evaluated. 
 
 3. Large-scale investigations on endangered species such as white dolphins can be conducted,making contributions to Taiwan＇s sustainable environment."

There are many drone companies in Taiwan, mainly based on hardware assembly. The competition is fierce with lower thresholds. The potential market to make money is big data analysis from drones. The core technology for building these coastal patrolenvironmental sustainability survey systems based on 5Gsmart drones is extremely small object detection. Many potential application fields can be developed from this core technology including: marine debris detection, marine oil pollution detection, white dolphin survey, security inspection system, coastalharbor patrol system, damaged ground pothole detection, illegal fishermen detection, landslide detection system,so on. .