中国农业期刊集成服务平台

当前位置: 首页 > 文章 > 基于TensorFlow的水族馆鱼类目标检测APP开发渔业现代化 2020 (2) 60-67

Position: Home > Articles > Aquarium fish target detection APP development based on TensorFlow Fishery Modernization 2020 (2) 60-67

基于TensorFlow的水族馆鱼类目标检测APP开发

作者：

张胜茂;刘洋;樊伟;邹国华;张衡;杨胜龙

单位：

中国水产科学研究院东海水产研究所

关键词：

水族馆;目标检测;TensorFlow;APP

摘要：

近年来深度学习在图像识别研究中取得突破进展,带动了目标检测技术的快速发展。利用目标检测技术开发水族馆鱼类目标检测APP,可以增强游客参观体验,提升科普效果。针对水族馆拍摄的80种鱼类,首先,使用LabelImg软件进行目标标记,再利用标记的目标导出成tfrecord数据;其次,选择ssd_mobilenet_v1模型进行数据训练,通过20万次的迭代训练获取到鱼类目标检测模型;最后,利用TensorFlow多目标检测API调用模型,定义2个接口和12个类,开发出Android系统手机APP。经过80种鱼类1 620张图片测试,正确率为92.59%,华为MHA-AL00手机目标检测平均时间40 ms。使用鱼类目标检测APP,能实现水族馆鱼类快速识别、多鱼类目标实时检测,可提升游客的参观体验,辅助科普量化评价。

译名：

Aquarium fish target detection APP development based on TensorFlow

作者：

ZHANG Shengmao;LIU Yang;FAN Wei;ZOU Guohua;ZHANG Heng;YANG Shenglong;Key Laboratory of East China Sea Fishery Resources Exploitation & Utilization,Ministry of Agriculture and Rural Affairs,East China Sea Fisheries Research Institute,Chinese Academy of Fishery Sciences;College of Information,Shanghai Ocean University;Shanghai Fishery Networking Technology Co.,Ltd.;Shanghai Junding Fishery Technology Co.,Ltd.;

关键词：

aquarium;;target detection;;TensorFlow;;APP

摘要：

In recent years, deep learning has made a breakthrough in image recognition research, which has led to the rapid development of target detection technology. It can enhance visitors' experience and improve the effect of science popularization by using target detection technology to develop aquarium fish target detection APP. In this paper, for 80 species of fish photographed at the aquarium, firstly, LabelImg software was used to mark the target, and the marked target was exported to generate tfrecord data. Secondly, ssd_mobilenet_v1 model was selected for data training, and the fish target detection model was obtained through 200 000 times of iterative training. Finally, 2 interfaces and 12 classes were defined with TensorFlow multi-targets detection API call model, and Android system mobile APP was developed. After testing 1 620 pictures of 80 species of fish, the accuracy rate was 92.59%. The average target detection time was 40 ms on Huawei MHA-AL00. Fish target detection APP can realize fast fish recognition in aquarium and real-time detection of multiple fish targets, improve visitors' experience and contribute to quantitative evaluation of science popularization.

相似文章

水族馆鱼类饲养管理技术 [刘明, 黄继成] 北京水产 1998 (4) 30-33
花胡椒鲷在日本培育初获成功 [殷禄阁] 水产科技情报 1982,000 (3) 31
水族馆内淡水鱼饲养池水质指标分析 [冯力霞, 王磊, 费玥, 栾钢] 水产学杂志 2013,26 (1) 27-31
大型水族馆的生命维持与警卫系统 [丁永良, 沈怡萱] 渔业现代化 1999 (2) 3-6
蓝鳃太阳鱼生物学特征及其应用前景 [胡逸农] 北京水产 1991,000 (2)
中国水族馆的可持续发展 [] 渔业现代化 2005,000 (2) 56
中华鲟水族馆驯养研究 [刘鉴毅, 张晓雁, 危起伟, 张辉, 杨道明, 朱永久, 陈细华, 杨德国, 郑跃平] 淡水渔业 2006,36 (5) 53-57
北京海洋馆大鲵科普展示工作介绍 [杨晓静, 陈微, 李灏, 郑东, 王磊, 管贺鹏, 王元群] 淡水渔业 2013,43 44-47
深海章鱼的孵卵周期超过4年 [杨林林] 渔业信息与战略 2014,29 (4) 316
中国水族馆的现状与发展 [王炜] 中国水产科学 1999,6 (3)

导出引用

计量

文章访问数: 9

HTML全文浏览量: 0

PDF下载量: 0

基于TensorFlow的水族馆鱼类目标检测APP开发

相似文章

所属期刊

渔业现代化

推荐期刊

淡水渔业

渔业现代化