人工智能识别与人工识别红外相机动物影像准确率分析:以上海大金山岛猕猴监测为例

doi:10.19741/j.issn.1673-4831.2022.0615

生态与农村环境学报 ›› 2023, Vol. 39 ›› Issue (7): 918-923.doi: 10.19741/j.issn.1673-4831.2022.0615

人工智能识别与人工识别红外相机动物影像准确率分析:以上海大金山岛猕猴监测为例

李必成¹, 张晨曦², 季钰翔¹, 孙锬锋³, 丁屹旻³, 张伟¹, 谢汉宾¹, 王军馥¹, 张云飞¹, 李雪梅¹, 王小明¹, 杨刚^1,2

1. 上海科技馆/上海科技馆长三角城市群生态安全与生物多样性保护实验室, 上海 200127;
2. 上海海洋大学海洋学院, 上海 201306;
3. 上海交通大学网络空间安全学院, 上海 200240

收稿日期:2022-06-17 出版日期:2023-07-25 发布日期:2023-07-19
通讯作者: 杨刚,E-mail:yangg@sstm.org.cn E-mail:yangg@sstm.org.cn
作者简介:李必成(1980-),男,安徽桐城人,副研究员,博士,研究方向为动物生态学和保护生物学。E-mail:libc@sstm.org.cn
基金资助:
上海市自然科学基金（20ZR1437100）；国家自然科学基金（31601872）；上海科技馆长三角城市群生态安全与生物多样性保护实验室项目

Analysis of the Accuracy of Artificial Intelligence Recognition and Artificial Recognition of Camera Traps Images: An Example of Macaca mulatta Monitoring on Dajinshan Island, Shanghai

LI Bi-cheng¹, ZHANG Chen-xi², JI Yu-xiang¹, SUN Tan-feng³, DING Yi-min³, ZHANG Wei¹, XIE Han-bin¹, WANG Jun-fu¹, ZHANG Yun-fei¹, LI Xue-mei¹, WANG Xiao-ming¹, YANG Gang^1,2

1. Shanghai Science and Technology Museum/Laboratory of Ecological Security and Biodiversity Conservation of Yangtze River Delta Urban Agglomeration, Shanghai Science and Technology Museum, Shanghai 200127, China;
2. College of Marine Science, Shanghai Ocean University, Shanghai 201306, China;
3. School of Cyber Science and Engineering, Shanghai Jiaotong University, Shanghai 200240, China

Received:2022-06-17 Online:2023-07-25 Published:2023-07-19

摘要/Abstract

摘要： 红外野生动物相机影像人工智能识别已成为生态学与人工智能交叉学科研究的热点之一。为探究人工智能识别红外相机动物影像的准确率及其影响因子，比较人工智能识别与人工识别的差异。以上海大金山岛猕猴（Macaca mulatta）监测为例，应用YOLO v3模型进行训练与测试，探讨YOLO v3模型识别大量红外相机图像的可行性。同时，对比人工智能图像识别与人工识别的准确率与识别效率，找出特定样本容量条件下识别方式的最优解。对11 106张照片的识别结果表明，人工智能识别总准确率为69.0%，均值为68.2%；人工识别总准确率为99.0%，均值为99.1%。人工识别准确率显著高于人工智能识别准确率（t=-9.256，df=22，P<0.01）。简单生境背景的人工智能识别准确率显著高于复杂生境背景（Z=-2.270，P=0.023）。简单生境背景的人工识别准确率与复杂生境背景无显著差异（Z=-0.406，P=0.685）。因此，人工智能识别更适用于生境及背景简单的红外影像，但需谨慎用于识别复杂生境的背景。同时，人工智能识别可用于对大量照片的初筛。人工识别可用于识别复杂生境背景的照片和对人工智能初筛后照片的复核。对于万张级的样本量，人工智能并未显示出明显的时间优势，人工识别反而具有准确率优势。随着各类训练数据集的不断建立与开放应用，对于大型脊椎动物，特别是一些公众熟知的明星物种的人工智能识别可能会率先代替人工识别。

关键词: 人工智能, 人工识别, 准确率, 猕猴, 大金山岛

Abstract: Artificial intelligence (AI) recognition of camera traps images has become one of the hot spots in the interdisciplinary research of ecology and AI. In order to explore the accuracy and influencing factors of artificial intelligence recognition of infrared camera animal images, the differences between artificial intelligence recognition and artificial recognition were compared. Taking the monitoring of macaques (Macaca mulatta) on Dajinshan Island in Shanghai as an example, the TOLO v3 model was applied for training and testing, and the feasibility of the TOLO v3 model to recognize a large number of infrared camera images was discussed. Meanwhile, the accuracy and recognition efficiency of AI image recognition and artificial recognition are compared to find out the optimal solution of recognition method under specific sample capacity. The recognition results of 11 106 photos show that the total recognition accuracy of artificial intelligence is 69.0%, and the average is 68.2%. The total accuracy of artificial recognition was 99.0%, and the average was 99.1%. The accuracy of artificial recognition was significantly higher than that of artificial intelligence (t=-9.256, df=22, P<0.01). The recognition accuracy of simple habitat background was significantly higher than that of complex habitat background (Z=-2.270, P=0.023). For artificial recognition, there was no significant difference in accuracy between simple habitat background and complex habitat background (Z=-0.406, P=0.685). AI recognition can be used for infrared images with a single habitat and background, but should be cautiously used for background recognition of complex habitats. In addition, AI recognition can be used for initial screening of large numbers of photographs. Artificial recognition can be used to identify photos with complex habitat backgrounds and to review photos after initial screening by artificial intelligence. For the sample size of ten thousand photos, artificial intelligence does not show obvious time advantage, but artificial recognition has an advantage of accuracy. With the continuous establishment and open application of various training data sets, artificial intelligence recognition for large vertebrates, especially for some well-known star species, may take the lead in replacing artificial recognition.

Key words: artificial intelligence, artificial recognition, accuracy, macaque, Dajinshan Island

中图分类号:

X835
Q958.1

李必成, 张晨曦, 季钰翔, 孙锬锋, 丁屹旻, 张伟, 谢汉宾, 王军馥, 张云飞, 李雪梅, 王小明, 杨刚. 人工智能识别与人工识别红外相机动物影像准确率分析:以上海大金山岛猕猴监测为例[J]. 生态与农村环境学报, 2023, 39(7): 918-923.

LI Bi-cheng, ZHANG Chen-xi, JI Yu-xiang, SUN Tan-feng, DING Yi-min, ZHANG Wei, XIE Han-bin, WANG Jun-fu, ZHANG Yun-fei, LI Xue-mei, WANG Xiao-ming, YANG Gang. Analysis of the Accuracy of Artificial Intelligence Recognition and Artificial Recognition of Camera Traps Images: An Example of Macaca mulatta Monitoring on Dajinshan Island, Shanghai[J]. Journal of Ecology and Rural Environment, 2023, 39(7): 918-923.

参考文献 28

[1]	MCSHEA W J, 申小莉, 刘芳, 等.中国的野生动物红外相机监测需要统一的标准[J].生物多样性, 2020, 28(9):1125-1131.[MCSHEA W J, SHEN Xiao-li, LIU Fang, et al.China's Wildlife Camera-trap Monitoring Needs a Unified Standard[J].Biodiversity Science, 2020, 28(9):1125-1131.]
[2]	肖治术, 李欣海, 姜广顺.红外相机技术在我国野生动物监测研究中的应用[J].生物多样性, 2014, 22(6):683-684.[XIAO Zhi-shu, LI Xin-hai, JIANG Guang-shun.Applications of Camera Trapping to Wildlife Surveys in China[J].Biodiversity Science, 2014, 22(6):683-684.]
[3]	朱淑怡, 段菲, 李晟.基于红外相机网络促进我国鸟类多样性监测:现状、问题与前景[J].生物多样性, 2017, 25(10):1114-1122.[ZHU Shu-yi, DUAN Fei, LI Sheng.Promoting Diversity Inventory and Monitoring of Birds through the Camera-trapping Network in China:Status, Challenges and Future Outlook[J].Biodiversity Science, 2017, 25(10):1114-1122.]
[4]	宫一男, 谭孟雨, 王震, 等.基于深度学习的红外相机动物影像人工智能识别:以东北虎豹国家公园为例[J].兽类学报, 2019, 39(4):458-465.[GONG Yi-nan, TAN Meng-yu, WANG Zhen, et al.AI Recognition of Infrared Camera Image of Wild Animals Based on Deep Learning:Northeast Tiger and Leopard National Park for Example[J].Acta Theriologica Sinica, 2019, 39(4):458-465.]
[5]	UENO M, KABATA R, HAYASHI H, et al.Automatic Individual Recognition of Japanese Macaques (Macaca fuscata) from Sequential Images[J].Ethology, 2022, 128(5):461-470.
[6]	李晟, 王大军, 卜红亮, 等.四川省老河沟自然保护区兽类多样性红外相机调查[J].兽类学报, 2016, 36(3):282-291.[LI Sheng, WANG Da-jun, BU Hong-liang, et al.Camera-trapping Survey on the Mammal Diversity of the Laohegou Nature Reserve, Sichuan Province[J].Acta Theriologica Sinica, 2016, 36(3):282-291.]
[7]	HINTON G E, SALAKHUTDINOV R R.Reducing the Dimensionality of Data with Neural Networks[J].Science, 2006, 313(5786):504-507.
[8]	SCHOFIELD D, NAGRANI A, ZISSERMAN A, et al.Chimpanzee Face Recognition from Videos in the Wild Using Deep Learning[J].Science Advances, 2019, 5(9):eaaw0736.
[9]	TANG X O, LIU K, CUI J Y, et al.IntentSearch:Capturing User Intention for One-click Internet Image Search[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(7):1342-1353.
[10]	EIKELBOOM J, WIND J, VAN DE VEN E.Improving the Precision and Accuracy of Animal Population Estimates with Aerial Image Object Detection[J].Methods in Ecology and Evolution, 2019, 10(11):1875-1887.
[11]	史春妹, 谢佳君, 顾佳音, 等.基于目标检测的东北虎个体自动识别[J].生态学报, 2021, 41(12):4685-4693.[SHI Chun-mei, XIE Jia-jun, GU Jia-yin, et al.Amur Tiger Individual Automatic Identification Based on Object Detection[J].Acta Ecologica Sinica, 2021, 41(12):4685-4693.]
[12]	谢将剑, 杨俊, 邢照亮, 等.多特征融合的鸟类物种识别方法[J].应用声学, 2020, 39(2):199-206.[XIE Jiang-jian, YANG Jun, XING Zhao-liang, et al.Bird Species Recognition Method Based on Multi-feature Fusion[J].Journal of Applied Acoustics, 2020, 39(2):199-206.]
[13]	TABAK M, NOROUZZADEH M, WOLFSON D.Machine Learning to Classify Animal Species in Camera Trap Images:Applications in Ecology[J].Methods in Ecology and Evolution, 2019, 10:585-590.
[14]	张伟, 谢汉宾, 杨刚, 等.上海大金山岛的两栖爬行动物[J].动物学杂志, 2021, 56(1):137-141.[ZHANG Wei, XIE Han-bin, YANG Gang, et al.Amphibians and Reptiles of the Dajinshan Island, Shanghai[J].Chinese Journal of Zoology, 2021, 56(1):137-141.]
[15]	沈志明, 李仲源, 王胜昌, 等.上海市大金山岛猕猴散养基地建立[J].上海实验动物科学, 1991, 11(1):20, 27.[SHEN Zhi-ming, LI Zhong-yuan, WANG Sheng-chang, et al.Establishment of Free-range Macaque Base in Dajinshan Island, Shanghai[J].Shanghai Laboratory Animal Science, 1991, 11(1):20, 27.]
[16]	MI Z J, JIANG X H, SUN T F, et al.GAN-generated Image Detection with Self-attention Mechanism against GAN Generator Defect[J].IEEE Journal of Selected Topics in Signal Processing, 2020, 14(5):969-981.
[17]	YANG G, XU J, WANG Y, et al.Evaluation of Microhabitats for Wild Birds in a Shanghai Urban Area Park[J].Urban Forestry & Urban Greening, 2015, 14(2):246-254.
[18]	李必成, 王军馥, 刘威, 等.上海绿地和农田生态系统陆生繁殖鸟类群落稳定性研究[J].生态与农村环境学报, 2020, 36(5):567-572.[LI Bi-cheng, WANG Jun-fu, LIU Wei, et al.Analyzing Community Stability of Terrestrial Breeding Birds in Greenspaces and Agricultural Ecosystems in Shanghai[J].Journal of Ecology and Rural Environment, 2020, 36(5):567-572.]
[19]	汤德俊.人脸识别中图像特征提取与匹配技术研究[D].大连:大连海事大学, 2013.[TANG De-jun.Research on Image Feature Extraction and Matching Technology in Face Recognition[D].Dalian:Dalian Maritime University, 2013.]
[20]	VALLETTA J J, TORNEY C, KINGS M, et al.Applications of Machine Learning in Animal Behaviour Studies[J].Animal Behaviour, 2017, 124:203-220.
[21]	STRACK R.AI Transforms Image Reconstruction[J].Nature Methods, 2018, 15(5):309.
[22]	朱林.基于特征加权与特征选择的数据挖掘算法研究[D].上海:上海交通大学, 2013.[ZHU Lin.Research on Feature Weighting and Feature Selection-based Data Mining Algorithms[D].Shanghai:Shanghai Jiao Tong University, 2013.]
[23]	李响, 谭南林, 李国正, 等.基于Zernike矩的人眼定位与状态识别[J].电子测量与仪器学报, 2015, 29(3):390-398.[LI Xiang, TAN Nan-lin, LI Guo-zheng, et al.Eye Location and Status Recognition Based on Zernike Moments[J].Journal of Electronic Measurement and Instrumentation, 2015, 29(3):390-398.]
[24]	曲培树, 董文会.基于眼睑曲率及模糊逻辑的人眼状态识别[J].计算机工程与科学, 2007, 29(8):50-51, 87.[QU Pei-shu, DONG Wen-hui.Eye States Recognition Based on Eyelid Curvature and Fuzzy Logic[J].Computer Engineering & Science, 2007, 29(8):50-51, 87.]
[25]	刘帝, 黄妍妍, 孟雪.一种基于人眼识别的疲劳驾驶检测算法[J].电子技术与软件工程, 2021(18):139-140.[LIU Di, HUANG Yan-yan, MENG Xue.A Fatigue Driving Detection Algorithm Based on Human Eye Recognition[J].Electronic Technology & Software Engineering, 2021(18):139-140.]
[26]	TABAK M A, NOROUZZADEH M S, WOLFSON D W, et al.Improving the Accessibility and Transferability of Machine Learning Algorithms for Identification of Animals in Camera Trap Images:MLWIC2[J].Ecology and Evolution, 2020, 10(19):10374-10383.
[27]	WITHAM C L.Automated Face Recognition of Rhesus Macaques[J].Journal of Neuroscience Methods, 2018, 300:157-165.
[28]	UENO M, YAMAMOTO H, YAMADA K, et al.Individual Recognition of Monkey (Macaca fuscata) and Human (Homo sapiens) Images in Primatologists[J].Journal of Comparative Psychology, 2021, 135(3):394-405.

人工智能识别与人工识别红外相机动物影像准确率分析:以上海大金山岛猕猴监测为例

Analysis of the Accuracy of Artificial Intelligence Recognition and Artificial Recognition of Camera Traps Images: An Example of Macaca mulatta Monitoring on Dajinshan Island, Shanghai

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献 28

相关文章 3

编辑推荐

Metrics

本文评价

[1]	罗鼎晖, 李翔, 骆蓓菁, 许洺山, 妥彬, 阎恩荣, 由文辉. 大金山岛常绿阔叶林和落叶阔叶林中小型土壤动物群落特征[J]. 生态与农村环境学报, 2020, 36(3): 349-357.
[2]	张福平, 高张, 李肖娟, 聂硕, 李玲, 刘洁遥. 基于最小数据集的周至县猕猴桃园地土壤质量评价[J]. 生态与农村环境学报, 2019, 35(1): 69-75.
[3]	秦卫华, 蒋明康, 王智, 刘鲁君. 河南太行山猕猴国家级自然保护区功能区调整与生态影响[J]. 生态与农村环境学报, 2005, 21(3): 78-80.