碩士 / 國立臺南大學 / 資訊工程學系碩士班 / 107 / In the aquaculture industry, everyone manages a large range of fish ponds. However, in recent years, the loss of labor in traditional industries has not only led to an increase average labor volume, but also affects the decline in output quality indirectly. Find convenient and simple management methods is becoming more and more important, and the demand for intelligent farming has come into being. In order to help the aquaculture manager manage the quality of fish and reduce the workload, we developed a preliminary method for fish tracking. The tracking method could be further applied to estimate the fish activity degree and the fish length. By feeding the information back to the manager, it is possible to reduce the workload of the manager and even to develop an automatic feeding system. The proposed tracking method combines the advantages of SSD (Single Shot Multibox Detector) and siamFC (Fully-Convolutional Siamese Networks) such that possesses good tracking performance and real-time processing speed.
Identifer | oai:union.ndltd.org:TW/107NTNT0392001 |
Date | January 2018 |
Creators | GUO, GUAN-YI, 郭冠毅 |
Contributors | Lee, Jiann-Shu, 李建樹 |
Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
Language | zh-TW |
Detected Language | English |
Type | 學位論文 ; thesis |
Format | 40 |
Page generated in 0.0017 seconds