用於識別絕育狀態影像的深度學習方法及其識別裝置

專利類型

發明

專利國別 (專利申請國家)

中華民國

專利申請案號

107124786

專利證號

I 682332

專利獲證名稱

用於識別絕育狀態影像的深度學習方法及其識別裝置

專利所屬機關 (申請機關)

國立彰化師範大學

獲證日期

2020/01/11

技術說明

本發明為一種用於醫療4.0之智慧辨識系統之影像辨識深度學習方法及識別裝置，可以應用於識別動物之絕育狀態影像。本發明運用人工智慧深度學習方法及識別裝置，以影像來辨識物件狀態，可應用於識別動物如貓、狗的絕育狀態。本發明可用於建立一辨識應用程式，該辨識應用程式識別出已結紮的動物影像，該深度學習方法透過一深度學習系統訓練影像，標示出包含生殖器官的感興趣區域，再經由一深度神經網路單元學習已結紮之生殖器官的特徵，並修正該辨識應用程式。輸入測試影像，並通過該辨識應用程式辨識該測試影像。在該辨識應用程式無法識別出呈現已結紮後之生殖器官的測試影像時，使呈現動物已結紮後之生殖器官的測試影像成為訓練影像，來持續訓練系統學習辨識。藉此，有效提升判斷動物是否結紮的效率與正確率。 The invention is an image recognition deep learning method and its identification device for the medical identification system of medical 4.0, which can be applied to identify images of sterilization status of animals. The invention uses the artificial intelligence deep learning method and the identification device to identify the state of an object using the image, and can be applied to identify the sterilization state of animals such as cats and dogs. The invention can be used to create an identification application that recognizes a ligated animal image. The deep learning method trains the image through a deep learning system, indicating the region of interest including the genital organs, and then passes through a deep neural network. The pathway unit learns the characteristics of the genital organs that have been ligated and corrects the identification application. The test image is input and the test image is identified by the identification application. When the identification application cannot recognize the test image of the genital organs after the ligation, the test image showing the ligated genital organs becomes a training image for the system learning and recognizing. Thereby, the efficiency and accuracy of system is effectively improved.

備註

連絡單位 (專責單位/部門名稱)

研究發展處

連絡電話

04-7232105轉1858

網址