A quality metric for object detection and focus for low-cost UAVs
Nadeesh Fernando, Manjusri Wickramasinghe, Kasun De Zoysa, Charitha Elvitigala · 2016
With the flooding of low-cost UAVs to the consumer market, there is a rejuvenated interest to develop applications using consumer UAVs to enhance the quality of day to day lives of us humans. However, the limited capacity with respect to memory and processing power makes adopting a consumer UAV to advanced tasks such as object detection and tracking challenging. The modern day object tracking quadcopters available in the market place is not concerned about maintaining the quality focus on the detected object. This in turn affects aspects that are dependent on the quality of the focus of the object detected such as in video streaming and tracking applications. This would result in less than sufficient experience for users of this technology. As a first step in solving the aforementioned issue, in this paper we propose a quality metric to measure the focus of the detected object in order to evaluate the quality of detections. This paves the way for advanced object tracking in quadcopters that are capable of preserving the quality of the object focus while tracking it.