欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2024, Vol. 40 ›› Issue (14): 157-164.doi: 10.11924/j.issn.1000-6850.casb2024-0057

• 水产·渔业 • 上一篇    

2个鲤群体(Cyprinus carpio L.)表型生长性状的AI测量与手工测量的相关性分析

马子尧1,2(), 潘红3, 王开阔1,2, 陈颖杰1,2, 曹逸铭2, 孙晓晴2, 张研2()   

  1. 1 上海海洋大学水产科学国家级实验教学示范中心，上海 201306
    2 农业农村部水生动物基因组学重点实验室/中国水产科学研究院生物技术研究中心/北京市渔业生物技术重点实验室，北京 100141
    3 北京市野生动物救护中心，北京 101300
  • 收稿日期:2024-01-19 修回日期:2024-03-21 出版日期:2024-05-15 发布日期:2024-05-09
  • 通讯作者:
    张研，女，1978年出生，副研究员，研究方向：水产遗传育种与生物技术。通信地址：100141 北京市丰台区永定路南青塔150号，E-mail：。
  • 作者简介:

    马子尧，男，1998年出生，硕士研究生，研究方向：鱼类遗传育种。通信地址：100141 北京市丰台区永定路南青塔150号，E-mail：。

  • 基金资助:
    国家重点研发计划农业生物种质资源挖掘与创新利用专项“主要单胃农业动物和水产生物优异种质资源精准鉴定”(2021YFD1200804); 中国水产科学研究院创新团队项目“水产生物遗传大数据研究及应用创新团队”(2023TD26)

Correlation Analysis Between AI and Manual Measurement for Morphological Traits of Two Cyprinus carpio L. Strains

MA Ziyao1,2(), PAN Hong3, WANG Kaikuo1,2, CHEN Yingjie1,2, CAO Yiming2, SUN Xiaoqing2, ZHANG Yan2()   

  1. 1 National Demonstration Center for Experimental Fisheries Science Education, Shanghai Ocean University, Shanghai 201306
    2 Key Laboratory of Aquatic Genomics, Ministry of Agriculture and Rural Affairs/ Beijing Key Laboratory of Fishery Biotechnology, Chinese Academy of Fishery Sciences, Beijing 100141
    3 Beijing Wildlife Rescue Center, Beijing 101300
  • Received:2024-01-19 Revised:2024-03-21 Published:2024-05-15 Online:2024-05-09

RichHTML

6

PDF (PC)

66

摘要：

鱼类形态是人工育种、功能基因定位以及水产养殖和生态学研究的重要种群资源。鲤(Cyprinus carpio L.)是中国重要的养殖淡水鱼类之一，鲤的自然和人工选择导致不同品种间的表型极具多样性。近年来，随着成像技术、计算能力和硬件设备的飞速发展，基于机器视觉的无损测量方法逐渐成为一种高效、可重复批量检测鱼体的方法。本研究旨在检验AI测量对鱼类表型性状测量的准确性。本文选择了表型指数差异显著的元江鲤(Cyprinus carpio yuankiang)和金背鲤(Cyprinus carpio var. Jinbei)2个群体，共计204尾鱼进行研究。针对全长、体长、体高、体厚、头长、尾柄长、尾柄高、吻长、眼径和眼间距共10种线性、圆形和空间性状进行手工测量和AI测量，并进行了比较分析。其结果显示：（1）元江鲤和金背鲤2个群体中，针对10种表型性状的测量，相较于手工测量，AI测量结果整体偏大；在表型指数中，体宽指数（体厚/体长），头长指数（头长/体长）和尾鳍指数（尾柄长/体长）2种测量方法均无显著差异，而2种方法在体深指数（体高/体长）结果中，呈现显著差异。（2）元江鲤和金背鲤群体中，对2种测量方法进行相关性和一致性分析，发现针对全长、体长、体高、体厚、头长和眼间距共6种性状，2种方法呈现高度一致性(r>0.85)；（3）2种测量方法重复性测量的一致性分析中，发现手工测量和AI测量均呈现较好的一致性(r>0.85)，但手工测量的一致性要略高于AI测量。综上所述，AI测量准确度的提升能够加快全长等线性表型性状测量的工作效率和选择育种速度，但对于圆形和空间等表型性状测量的准确度，还需进一步提升。

关键词: 手工测量, AI测量, 元江鲤, 金背鲤

Abstract:

The morphology of fish is an important population resource for artificial breeding, functional gene localization, as well as aquaculture and ecological research. The common carp (Cyprinus carpio L.) is one of the important freshwater fish species farmed in China. Natural and artificial selection have resulted in a great diversity of phenotypes among different varieties of carp. In recent years, with the rapid development of imaging technology, computational capabilities, and hardware devices, non-destructive measurement methods based on machine vision have rapidly emerged, making it an efficient and repeatable method for batch inspection of fish bodies. In order to test the accuracy of AI measurement for fish phenotypic traits, manual measurement and AI measurement were used in two ways, respectively, for the Cyprinus carpio yuankiang and the Cyprinus carpio var. Jinbei, which have significant differences in phenotypic indices. A total of 204 fish from two carp populations were selected, and 10 linear, circular, and spatial traits including total length, body length, body depth, body thickness, head length, caudal peduncle length, caudal peduncle depth, snout length, eye diameter and interorbital width were measured and compared. The results show: (1) In the two populations of Cyprinus carpio yuankiang and Cyprinus carpio var. Jinbei, for the measurement of 10 phenotypic traits, compared to manual measurement, the AI measurement results are generally larger. In the phenotypic index, there is no significant difference between the two measurement methods in the body width index (body thickness/body length), head length index (head length/body length), and caudal fin index (caudal peduncle length/body length), while the two methods show significant differences in the results of the body depth index (body depth/body length). (2) In the populations of Cyprinus carpio yuankiang and Cyprinus carpio var. Jinbei, a correlation and consistency analysis of the two measurement methods found that for a total of 6 traits including total length, body length, body depth, body thickness, head length, and interorbital width, the two methods show high consistency (r>0.85). (3) In the consistency analysis of the repeatability measurement of the two measurement methods, it was found that both manual measurement and AI measurement show good consistency (r>0.85), but the consistency of manual measurement is slightly higher than that of AI measurement. Overall, the improvement of the accuracy of AI measurement can speed up the work efficiency of measuring linear phenotypic traits such as total length and the speed of selective breeding, but the accuracy of measuring circular and spatial phenotypic traits needs to be further improved.

Key words: manual measurement, AI measurement, Cyprinus carpio yuankiang, Cyprinus carpio var. Jinbei