作者：王伟民，程进，邹小平，朴林华
单位：北京信息科技大学北京市传感器重点实验室，北京 100101
中图分类号：TP183；TP391.4        
文献标识码：A        
文章编号：1006-883X(2022)02-0011-05
收稿日期：2021-09-23

摘要：文章研究了一种不同于其他产品的新型宠物自动投食器——基于卷积神经网络的猫狗识别自动投食器。与市面上其他投喂对象单一的宠物自动投食器相比，该款投食器能够根据图像识别结果去投喂多种类型的宠物，例如猫、狗。该款投食器将来能够运用在宠物家庭、宠物店、宠物主题餐厅等猫狗混养的场所。该装置使用的技术主要涉及机器学习的相关内容，通过训练和搭建卷积神经网络来准确识别猫、狗的图像，并且根据图像识别结果投喂相应的猫粮或者狗粮。同时，利用安装在机身上的红外传感器作为投喂开关，负责生物接近检测，以判断投食器是否需要进行投喂，从而使整个投食过程更加合理与智能。
关键词：宠物自动投食器；猫狗混养；机器学习；卷积神经网络；生物接近检测

Cat and Dog Recognition Automatic Feeder Based on Convolutional Neural Network
WANG Weimin, CHENG Jin, ZOU Xiaoping, PIAO Linhua
(Beijing Sensor Key Laboratory, Beijing Information Science & Technology University, Beijing 100101, China)
Abstract: The paper studies a new pet automatic feeder called cat and dog recognition automatic feeder based on convolutional neural network, which is different from other products. Compared with other pet automatic feeder which has a single feeding object in the market, the feeder can feed multiple types of pets according to the image recognition results, such as cats and dogs. In addition, the device can be used in pet homes, pet stores, pet themed restaurants and other places with mixed breeding of cats and dogs. The technology used in the device mainly involves machine learning. Through training and building convolutional neural network, it can accurately identify cat and dog images, and feed corresponding cat or dog food according to the image recognition results. At the same time, the infrared sensor installed on the fuselage is used as the feeding switch, responsible for biological proximity detection to judge whether the feeding device needs to be feeding, so as to make the whole feeding process more reasonable and intelligent.
Key words: pet automatic feeder; mixed breeding of cats and dogs; machine learning; convolutional neural network; biological proximity detection

阅读全文

备注：2022年 第28卷 第02期