2022
06/30
相关创新主体

创新背景

数字技术和人工智能时代,对人类的行为、面部情绪识别技术已经在不断发展,关于动物的研究还需要继续挖掘。动物保护和研究需要对动物的行为以及传达出来的情绪进行辨别,才能更好地进行动物福利机制建立和生物多样性保护。

 

创新过程

2021年,荷兰赫宁根大学的Suresh Neethirajan博士带领团队在世界范围内搜集数万头农场动物的数据,对它们的面部特征和变化进行研究分析,对与动物情绪相关的数据进行追踪优化。

得到十三种面部活动和九种情绪状态,提出了一种基于 YoloV3、Faster YoloV4 的图像实时识别系统和集成卷积神经网络,对农场动物表情和情绪状态进行检测。之后基于 Python 的算法,创建了一个能够用于识别农场动物情绪的人工智能系统,利用面部识别技术跟踪动物状态,帮助农场进行精细化管理,加强人与动物的和谐交流,推动动物感知领域技术突破。

对于动物面部情绪的识别有助于畜牧业可持续高质量健康发展,完善动物福利审查机制。同样,对动物行为进行追踪检测也有同样的积极效果。

2022年,瑞士苏黎世联邦理工学院与苏黎世大学的研究团队开发出一种追踪分析动物行为的人工智能方法,可以对动物活动和互动行为进行自动录像和分析,减少观察实验耗费的人力和精力。研究结果已被发表在4月的《自然·机器智能》期刊上。

研究开发出高灵敏度的图像分析算法,利用计算机视觉技术和机器学习区分单个动物,识别动物在一段时间内的特定行为和行为变化,包括与好奇、恐惧、和睦相关的各种人眼难以察觉的行为和变化。利用人工智能进行自动化观察分析,捕捉动物复杂环境下的社会行为,大大减少研究人员观看录像的同时,令研究结果依靠同样的标准更容易对比。苏黎世动物园将该算法用于动物饲养与行为研究。

创新价值

创新人工智能算法识别动物情绪并追踪分析动物行为,提高了人类对动物的了解,并且可以对动物患病和异常行为做出预测监控,降低风险损失,改善动物福利,促进人与动物、人与自然和谐相处。

创新关键点

创新人工智能系统算法,捕捉动物情绪和行为,帮助了解动物心理。

 

Artificial intelligence algorithms help track animal behavior and recognize emotions

In 2021, Dr. Suresh Neethirajan of the University of Heningen in the Netherlands will lead a team to collect data on tens of thousands of farm animals around the world, conduct research and analysis on their facial features and changes, and track and optimize data related to animal emotions.
Thirteen facial activities and nine emotional states were obtained, and a real-time image recognition system based on YoloV3 and Faster YoloV4 and an integrated convolutional neural network were proposed to detect the expressions and emotional states of farm animals. Afterwards, based on the Python algorithm, an artificial intelligence system that can be used to identify the emotions of farm animals was created, using facial recognition technology to track animal status, helping farms to carry out refined management, strengthening the harmonious communication between humans and animals, and promoting technological breakthroughs in the field of animal perception .
Recognition of animal facial emotions contributes to the sustainable, high-quality and healthy development of animal husbandry and improves the animal welfare review mechanism. Likewise, tracking animal behavior has the same positive effect.
In 2022, a research team from the Swiss Federal Institute of Technology Zurich and the University of Zurich developed an artificial intelligence method for tracking and analyzing animal behavior, which can automatically record and analyze animal activities and interactive behaviors, reducing the manpower and energy consumed by observation experiments. The findings have been published in the April issue of the journal Nature Machine Intelligence.
Research has developed high-sensitivity image analysis algorithms that use computer vision technology and machine learning to distinguish individual animals and identify specific animal behaviors and behavioral changes over a period of time, including various types of curiosity, fear, and harmony that are imperceptible to the human eye. behavior and change.
Using artificial intelligence to perform automated observation and analysis to capture the social behavior of animals in complex environments, greatly reduces the need for researchers to watch videos while making research results easier to compare based on the same standards. Zurich Zoo uses the algorithm for animal husbandry and behavior research.
Innovative artificial intelligence algorithms identify animal emotions and track and analyze animal behavior, which improves human understanding of animals, and can predict and monitor animal diseases and abnormal behavior, reduce risk losses, improve animal welfare, and promote human-animal, human-to-animal and human-to-animal Live in harmony with nature.

智能推荐

  • 人工智能+古文字学 | 人工智能帮助研究古文字文本

    2022-07-28

    将人工智能运用于古文字学领域,帮助重建古老的文字文本,探索人类的历史文明。

    涉及学科
    涉及领域
    关键词 
  • AI+农业 | 利用人工智能预测农作物生长状况

    2022-08-17

    利用人工智能收集农业作物生长收获全过程的相关数据,预测气候变化对农业作物的影响,最大限度地对作物生长进行合适干预,帮助增加农业作物收成,保障粮食供应。

    涉及学科
    涉及领域
    关键词 
  • 利用人工智能预测蛋白质3D结构

    2022-08-05

    根据DeepMind和伦敦大学学院(UCL)研究人员的一项研究,深度学习系统可以利用蛋白质的基因序列,比以往任何建模系统都更准确地预测蛋白质的结构。

    涉及学科
    涉及领域
    关键词 
  • 通过自监督学习实验发现AI模型与人脑的相似性

    2022-07-28

    Meta AI、美国哥伦比亚大学、多伦多大学等的研究人员完成了一个关于深度学习模型和人脑之间相似性的研究。研究发现,AI模型Wav2Vec 2.0与人类大脑处理语音的方式非常相似,甚至AI也像人类一样,对“母语”有更强的辨别能力,如法语模型就比英语模型更容易感知来自法语的刺激。

    涉及学科
    涉及领域
    关键词