農業部-水產試驗所全球資訊網農業部-水產試驗所全球資訊網

本計畫建立於107年度建立之魚類辨識系統的架構上，整合設計一個網頁雲端架構系 統、Socket通訊介面與魚類辨識系統，讓雲端魚類辨識系統的辨識速度可以在使用 者感受度較好的情況下完成辨識，藉由這個系統希望可以讓使用者可以快速且簡易 的識別魚的種類，並且透過使用者收集更多魚類相關照片，未來可以再增加訓練資 料庫的模型以提高辨識率，目前的辨識率雖然高達88%，還是有少數樣本需要再加以 修正可以讓整體系統更加完整，最後讓實際讓使用者體驗雲端魚類辨識平臺，讓此 系統可以提升日常生活中大家對魚類的認知。

The study is based on the structure of the fish identification system established in the year of 107. It integrates the design of a web cloud architecture system, Socket communication interface and fish identification system, so that the recognition speed of the cloud fish identification system can be better in the user's perception. The identification is completed. With this system, the user can quickly and easily identify the type of fish, and collect more fish-related photos through the user. In the future, the model of the training database can be increased to improve the recognition rate. Although the recognition rate is as high as 88%, there are still a few samples that need to be revised to make the overall system more complete. Finally, let the user experience the cloud fish identification platform, so that the system can enhance the awareness of fish in daily life.