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
|
摘要
为解决大量红外相机监测影像数据量庞大、亟待快速和自动识别的问题,本研究以东北虎豹国家公园为例,应用卷积神经网络,通过深度学习算法对红外相机影像实现物种自动识别。本研究选择8个物种的红外相机视频影像,以50帧率均匀采集成图片格式,每个物种筛选不同角度、不同环境条件的图片,建立图片数据集,包括训练集2 074张,测试集519张。对图片进行目标打框、类别标注,选用darknet框架下的YOLO v3模型进行训练。首先不区分昼(RGB)夜(灰度)图像进行训练,再区分昼夜进行训练,最后分别对昼夜图像利用微调(fine-tune)进行训练。研究初步结果显示,基于YOLO v3模型对自然条件下拍摄的红外相机图像进行物种自动识别能够一定程度减轻人力负担,但其效果还需通过完善数据集进行提升。fine-tune在小数据集时或可作为辅助。模型对8个物种识别的平均精确率达到84.9%～96.0%,且模型收敛。
Video data of wild animals from infrared cameras always has a large quantity,which takes a lot of work to select and identify.In order to meet the demand of fast automatic identification,this study,using Northeast Tiger and Leopard National Park as an example,is to explore the practicability of using convolutional neural networks,deep learning to automatically identify different animal species,using videos taken by infrared cameras in the wild,under natural conditions.Pictures of each 8 species,captured from the videos from different seasons and of different conditions,consist of the data set 2074 pictures for train set and 519 for test set.Region of Interest is selected and labeled,the model is YOLO v3 under darknet framework.All pictures are in one data set in the first group of experiment.In the second group,pictures are divided into day(RGB)and night(Gray),and in the third group,divided into day(RGB)and night(Gray)while finetuning is used.The mean average precision of our models is from 84.9% to 96.0%,and the models converge.Results show that althoug it is still needed to use better train set to improve the models,using YOLO v3 to identify wild animals automatically is practicable to save manpower and fine-tuning could be an assistance when the train set is small.
Video data of wild animals from infrared cameras always has a large quantity,which takes a lot of work to select and identify.In order to meet the demand of fast automatic identification,this study,using Northeast Tiger and Leopard National Park as an example,is to explore the practicability of using convolutional neural networks,deep learning to automatically identify different animal species,using videos taken by infrared cameras in the wild,under natural conditions.Pictures of each 8 species,captured from the videos from different seasons and of different conditions,consist of the data set 2074 pictures for train set and 519 for test set.Region of Interest is selected and labeled,the model is YOLO v3 under darknet framework.All pictures are in one data set in the first group of experiment.In the second group,pictures are divided into day(RGB)and night(Gray),and in the third group,divided into day(RGB)and night(Gray)while finetuning is used.The mean average precision of our models is from 84.9% to 96.0%,and the models converge.Results show that althoug it is still needed to use better train set to improve the models,using YOLO v3 to identify wild animals automatically is practicable to save manpower and fine-tuning could be an assistance when the train set is small.
引文
Ahumada J A,Silva C E F,Gajapersad K,Hallam C,Hurtado J,Martin E,McWilliam A,Mugerwa B,0'Brien T,Rovero F,Sheil D,Spironello W R,Winani N,Andelman S J.2011.Community structure and diversity of tropical forest mammals:data from a global camera trap network.Philosophical Transactions of the Royal Society B:Biological Sciences,366(1578):2703-271 1.
Browning E,Bolton M,Owen E,Shoji A,Guilford T,Freeman R.2017,Predicting animal behaviour using deep learning:gps data alone accurately predict diving in seabirds.Methods in Ecology and Evolution,9:681-692
Chen C,Qi F.2019.Review on development of convolutional neural network and its application in computer vision.Computer Science,46(3):63-73.(in Chinese)
Chen J Y,Nasendelger,Sun Q H,Zhang L J,Tang J R,Lang J M,Liu T,Liu K P,Xiao W H,Bao W D.2011.Amur tiger and prey in Jilin Hunchun National Nature Reserve,China.Chinese Journal of Zoology,46(2):46-52.(in Chinese)
Ding W,Taylor G.2016.Automatic moth detection from trap images for pest management.Computers and Electronics in Agriculture, 123:17-28.
Felzenszwalb P F,Girshick R B,Mcallester D,Ramanan D.2010.Object detection with discriminatively trained part-based models.IEEE Transactions on Software Engineering,32(9):1627-1645.
Girshick R,Donahue J,Darrelland T,Malik J.2014.Rich feature hierarchies for object detection and semantic segmentation.2014 IEEE Conference on Computer Vision and Pattern Recognition.IEEE.
Gomez A,Salazar A,Vargas F.2016.Towards automatic wild animal monitoring:identification of animal species in camera-trap images using very deep convolutional neural networks.Ecological Informatics,41:24-32.
Guo Q X,Wang H R,Gao M X.2008.Simulation on the effect of landscape fragmentation on population dynamics of the main prey of Amur tiger.Acta Ecologica Sinica,30(1):15-23.(in Chinese)
Konovalov D A,Hillcoat S,Williams G,Birtles R A,Gardiner N,Curnock M I.2018.Individual minke whale recognition using deep learning convolutional neural networks.Journal of Geoscience and Environment Protection,6:25-36.
Li F.2015.The research on automatic identification of butterfly species based on the digital image.Ph.D thesis.Beijing:Beijing Forestry University.(in Chinese)
Li S,Wang D J,Xiao Z S,LU X H,Wang T M,Feng L M,Wang Y.2014.Camera-trapping in wildlife research and conservation in China:review and outlook.Biodiversity Science, 22(6):685-695.(in Chinese)
Liu W D,Li A Q,Zhang J G,Xie J J,Bao W D.2018.Automatic identification method for terrestrial wildlife in Saihanwula National Nature Reserve in Inner Mongolia of northern China based on ROI-CNN.Journal of Beijing Forestry University,40(08):127-135.(in Chinese)
Norouzzadeh M S,Nguyen A,Kosmala M,Swanson A,Packer C,Clune J.2017.Automatically identify:ing wild animals in camera trap images with deep learning.Proceedings of the National Academy of Scien.ces,115(25):E5716-E5725.
Redmon J,Divvala S,Girshick R,Farhadi A.2015.You only look once:unified,real-time object detection.DOI:10.1109/CVPR.2016.91.
Redmon J,Farhadi A.2018.YOLO v3:an incremental improvement
Redmon J,Farhadi A.2017.[ieee 2017 ieee conference on computer vision and pattern recognition(cvpr)-honolulu,hi(2017.7.21-2017.7.26)]2017 ieee conference on computer vision and pattern recognition(cvpr)-YOLO 9000:better,faster,stronger.6517-6525.
Tajbakhsh N,Shin J Y,Gurudu S R,Hurst R,Kendall C B,Gotway M B,Liang J M.2016.Convolutional neural networks for medical image analysis:fine tuning or full training? IEEE Transactions on Medical Imaging,35(5):1299-1306.
Wang G H,Li S Q,Shi Z P,Wang S N,Ye J P,Zhou Q H.2016.Preliminary survey of mammal and bird diversity of Guangxi Maoershan National Nature Reserve-Based on infrared camera monitoring.Acta Theriologica Sinica,36(3):338-347.(in Chinese)
Xiao W H,Feng L M,Zhao X D,Yang H T,Dou H L,Cheng Y C,Mu P,Wang T M,Ge J P.2014.Distribution and abundance of Amur tiger,Amur leopard and their ungulate prey in Hunchun National Nature Reserve,Jilin.Biodiversity Science, 22(6):717-724.(in Chinese)
Xiao W H.2014.Amur tiger(Panthera tigris altaica)and tis prey in Hunchun Nature Reserve,jilin,China:their population size,distribution and occupancy.Ph.D thesis,Beijing:Beijing Normal University.(in Chinese).
Xie J Y,Hou Q,Shi Y H,Lv P,Jing L P,Zhuang F Z,Zhang J P,Tan X Y,Xu S Q.2018.The automatic identification of butterfly species.Journal of Computer Research and Development,55(08):1609-1618.
Yang H T.2018.Population ecology and coexistence mechanism of Amur tiger and Amur leopard,Northeast China Ph.D thesis.Beijing:Beijing Normal University.(in Chinese)
Zhang C Z.2013.Study on Amur tiger and prey monitoring,and conflict between tiger and human in Northeast China.Ph.D thesis.Harbin:Northeast Forestry University).(in Chinese)
Zhou A M,Ma P P,Xi T Y,Wang J N,Feng J,Shao Z Z,Tao Y L,Yao Q.2017.Automatic identification of butterfly specimen images at the family level based on deep learning method.Acta Entomologica Sinica,60(11):1339-1348.(in Chinese)
Zhou F Y,Jin L P,Dong J.2017.Review of convolutional neural network.Chinese Journal of Computers,40(6):1229-1251.(in Chinese)
刘文定,李安琪,张军国,谢将剑,鲍伟东.2018·基于ROI-CNN的赛罕乌拉国家级自然保护区陆生野生动物自动识别.北京林业大学学报,40(8):127-135.
李凡.2015.基于数字图像的蝴蝶种类自动识别研究.北京林业大学博士学位论文.
李晟,王大军,肖治术,李欣海,王天明,冯利民,王云. 2014.红外相机技术在我国野生动物研究与保护中的应用与前景.生物多样性,22(6):685-695.
汪国海,李生强,施泽攀,王绍能,叶建平,周岐海.2016.广西猫儿山自然保护区的兽类和鸟类多样性初步调查——基于红外相机监测数据.兽类学报,36(3):338-347.
张常智.2013.东北虎及其猎物种群监测和人虎冲突研究.东北林业大学博士学文论文.
陈九屹,那顺得力格尔,孙全辉,张逦嘉,唐继荣,郎建民,刘通,刘昆鹏,肖文宏,鲍伟东.2011.吉林珲春自然保护区东北虎及其猎物资源调查.动物学杂志,46(2)46-52.
陈超,齐峰.2019.卷积神经网络的发展及其在计算机视觉领域中的应用综述.计算机科学,46(3):63-73.
肖文宏.2014.东北虎(Panthera tigris altaica)与猎物的种群分布、数量和占据研究.北京:北京师范大学博士学论文.
肖文宏,冯利民,赵小丹,杨海涛,窦海龙,程艳超,牟溥,王天明,葛剑平.2014.吉林珲春自然保护区东北虎和东北豹及其有蹄类猎物的多度与分布.生物多样性,22(6):717-724.
国庆喜,王化儒,高梅香.2008.景观破碎化对东北虎主要猎物种群动态影响的模拟.生态学报,30(1):15-23.
周飞燕,金林鹏,董军.2017.卷积神经网络研究综述.计算机学报,40(6):1229-1251.
周爱明,马鹏鹏,席天宇,王江宁,冯晋,邵泽中,陶玉磊,姚青.2017.基于深度学习的蝴蝶科级标本图像自动识别.昆虫学报,60(11):1339-1348.
杨海涛.2018.东北虎东北豹种群生态学和共存机制研究.北京:北京师范大学博士学文论文.
谢娟英,侯琦,史颖欢,吕鹏,景丽萍,庄福振,张军平,谭晓阳,许升全.2018.蝴蝶种类自动识别研究.计算机研究与发展,55(8):1609-1618.
Browning E,Bolton M,Owen E,Shoji A,Guilford T,Freeman R.2017,Predicting animal behaviour using deep learning:gps data alone accurately predict diving in seabirds.Methods in Ecology and Evolution,9:681-692
Chen C,Qi F.2019.Review on development of convolutional neural network and its application in computer vision.Computer Science,46(3):63-73.(in Chinese)
Chen J Y,Nasendelger,Sun Q H,Zhang L J,Tang J R,Lang J M,Liu T,Liu K P,Xiao W H,Bao W D.2011.Amur tiger and prey in Jilin Hunchun National Nature Reserve,China.Chinese Journal of Zoology,46(2):46-52.(in Chinese)
Ding W,Taylor G.2016.Automatic moth detection from trap images for pest management.Computers and Electronics in Agriculture, 123:17-28.
Felzenszwalb P F,Girshick R B,Mcallester D,Ramanan D.2010.Object detection with discriminatively trained part-based models.IEEE Transactions on Software Engineering,32(9):1627-1645.
Girshick R,Donahue J,Darrelland T,Malik J.2014.Rich feature hierarchies for object detection and semantic segmentation.2014 IEEE Conference on Computer Vision and Pattern Recognition.IEEE.
Gomez A,Salazar A,Vargas F.2016.Towards automatic wild animal monitoring:identification of animal species in camera-trap images using very deep convolutional neural networks.Ecological Informatics,41:24-32.
Guo Q X,Wang H R,Gao M X.2008.Simulation on the effect of landscape fragmentation on population dynamics of the main prey of Amur tiger.Acta Ecologica Sinica,30(1):15-23.(in Chinese)
Konovalov D A,Hillcoat S,Williams G,Birtles R A,Gardiner N,Curnock M I.2018.Individual minke whale recognition using deep learning convolutional neural networks.Journal of Geoscience and Environment Protection,6:25-36.
Li F.2015.The research on automatic identification of butterfly species based on the digital image.Ph.D thesis.Beijing:Beijing Forestry University.(in Chinese)
Li S,Wang D J,Xiao Z S,LU X H,Wang T M,Feng L M,Wang Y.2014.Camera-trapping in wildlife research and conservation in China:review and outlook.Biodiversity Science, 22(6):685-695.(in Chinese)
Liu W D,Li A Q,Zhang J G,Xie J J,Bao W D.2018.Automatic identification method for terrestrial wildlife in Saihanwula National Nature Reserve in Inner Mongolia of northern China based on ROI-CNN.Journal of Beijing Forestry University,40(08):127-135.(in Chinese)
Norouzzadeh M S,Nguyen A,Kosmala M,Swanson A,Packer C,Clune J.2017.Automatically identify:ing wild animals in camera trap images with deep learning.Proceedings of the National Academy of Scien.ces,115(25):E5716-E5725.
Redmon J,Divvala S,Girshick R,Farhadi A.2015.You only look once:unified,real-time object detection.DOI:10.1109/CVPR.2016.91.
Redmon J,Farhadi A.2018.YOLO v3:an incremental improvement
Redmon J,Farhadi A.2017.[ieee 2017 ieee conference on computer vision and pattern recognition(cvpr)-honolulu,hi(2017.7.21-2017.7.26)]2017 ieee conference on computer vision and pattern recognition(cvpr)-YOLO 9000:better,faster,stronger.6517-6525.
Tajbakhsh N,Shin J Y,Gurudu S R,Hurst R,Kendall C B,Gotway M B,Liang J M.2016.Convolutional neural networks for medical image analysis:fine tuning or full training? IEEE Transactions on Medical Imaging,35(5):1299-1306.
Wang G H,Li S Q,Shi Z P,Wang S N,Ye J P,Zhou Q H.2016.Preliminary survey of mammal and bird diversity of Guangxi Maoershan National Nature Reserve-Based on infrared camera monitoring.Acta Theriologica Sinica,36(3):338-347.(in Chinese)
Xiao W H,Feng L M,Zhao X D,Yang H T,Dou H L,Cheng Y C,Mu P,Wang T M,Ge J P.2014.Distribution and abundance of Amur tiger,Amur leopard and their ungulate prey in Hunchun National Nature Reserve,Jilin.Biodiversity Science, 22(6):717-724.(in Chinese)
Xiao W H.2014.Amur tiger(Panthera tigris altaica)and tis prey in Hunchun Nature Reserve,jilin,China:their population size,distribution and occupancy.Ph.D thesis,Beijing:Beijing Normal University.(in Chinese).
Xie J Y,Hou Q,Shi Y H,Lv P,Jing L P,Zhuang F Z,Zhang J P,Tan X Y,Xu S Q.2018.The automatic identification of butterfly species.Journal of Computer Research and Development,55(08):1609-1618.
Yang H T.2018.Population ecology and coexistence mechanism of Amur tiger and Amur leopard,Northeast China Ph.D thesis.Beijing:Beijing Normal University.(in Chinese)
Zhang C Z.2013.Study on Amur tiger and prey monitoring,and conflict between tiger and human in Northeast China.Ph.D thesis.Harbin:Northeast Forestry University).(in Chinese)
Zhou A M,Ma P P,Xi T Y,Wang J N,Feng J,Shao Z Z,Tao Y L,Yao Q.2017.Automatic identification of butterfly specimen images at the family level based on deep learning method.Acta Entomologica Sinica,60(11):1339-1348.(in Chinese)
Zhou F Y,Jin L P,Dong J.2017.Review of convolutional neural network.Chinese Journal of Computers,40(6):1229-1251.(in Chinese)
刘文定,李安琪,张军国,谢将剑,鲍伟东.2018·基于ROI-CNN的赛罕乌拉国家级自然保护区陆生野生动物自动识别.北京林业大学学报,40(8):127-135.
李凡.2015.基于数字图像的蝴蝶种类自动识别研究.北京林业大学博士学位论文.
李晟,王大军,肖治术,李欣海,王天明,冯利民,王云. 2014.红外相机技术在我国野生动物研究与保护中的应用与前景.生物多样性,22(6):685-695.
汪国海,李生强,施泽攀,王绍能,叶建平,周岐海.2016.广西猫儿山自然保护区的兽类和鸟类多样性初步调查——基于红外相机监测数据.兽类学报,36(3):338-347.
张常智.2013.东北虎及其猎物种群监测和人虎冲突研究.东北林业大学博士学文论文.
陈九屹,那顺得力格尔,孙全辉,张逦嘉,唐继荣,郎建民,刘通,刘昆鹏,肖文宏,鲍伟东.2011.吉林珲春自然保护区东北虎及其猎物资源调查.动物学杂志,46(2)46-52.
陈超,齐峰.2019.卷积神经网络的发展及其在计算机视觉领域中的应用综述.计算机科学,46(3):63-73.
肖文宏.2014.东北虎(Panthera tigris altaica)与猎物的种群分布、数量和占据研究.北京:北京师范大学博士学论文.
肖文宏,冯利民,赵小丹,杨海涛,窦海龙,程艳超,牟溥,王天明,葛剑平.2014.吉林珲春自然保护区东北虎和东北豹及其有蹄类猎物的多度与分布.生物多样性,22(6):717-724.
国庆喜,王化儒,高梅香.2008.景观破碎化对东北虎主要猎物种群动态影响的模拟.生态学报,30(1):15-23.
周飞燕,金林鹏,董军.2017.卷积神经网络研究综述.计算机学报,40(6):1229-1251.
周爱明,马鹏鹏,席天宇,王江宁,冯晋,邵泽中,陶玉磊,姚青.2017.基于深度学习的蝴蝶科级标本图像自动识别.昆虫学报,60(11):1339-1348.
杨海涛.2018.东北虎东北豹种群生态学和共存机制研究.北京:北京师范大学博士学文论文.
谢娟英,侯琦,史颖欢,吕鹏,景丽萍,庄福振,张军平,谭晓阳,许升全.2018.蝴蝶种类自动识别研究.计算机研究与发展,55(8):1609-1618.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |