Automated tracking of monkey groups: recognition of social structure and behavior

 描述(申请人提供)：描述群居恒河猴的社会联系和等级关系的模式是非常耗费人力的。我们建议开发和实施创新的软件和硬件技术，这些技术将自动连续收集复杂物种-典型的恒河猴社会群体中的社交互动数据。拟议的系统将持续和同时跟踪在一大群猴子的半英亩生活区中多个个体的精确3D位置。我们将设计和制造舒适的项圈来识别个人身份，包括持续广播每种动物唯一代码的有源RFID标签。大院周边内和周围的RFID探测器可以三角测量每只动物的精确位置，精确到25厘米以内，以及它们的移动方向和速度。我们将跟踪构成社会等级核心的成年男性和女性(总共45人)。个体之间的距离是推断社会联系的关键诊断，因为猴子之间的距离很少意外地出现在彼此之间25厘米的地方。因此，知道在该距离上花费的时间量可以推断从属关系。从属关系、回避和接近时间的模式将允许推断群体的社会等级。结合社会网络分析，我们将确定群体关联和社会不稳定的模式。此外，我们将开发一个部署(在第三年和第四年)的RFID引导相机系统，该系统可以自动记录预定目标对象的高分辨率视频，而无需人工干预，从而大大提高了从这些视频片段进行后续人类行为编码的效率。在该项目下将开发三个软件系统：1)行为推理软件，从自动跟踪数据推断从属、等级、攻击性和性行为；2)社交网络分析软件，用于识别社会不稳定；3)视频识别软件，将从由RFID引导的摄像头收集的视频片段中自动编码社会行为。这种硬件和软件的结合将提供一个增强的动物资源，使从药物到神经侮辱等各种操作的社会影响的调查成为可能。此外，该系统还可用于监测群体健康和识别个体疾病行为。这种动物资源将促进由NIMH、NIDA或NICHD资助的研究，并支持和加强灵长类群体的管理。

 DESCRIPTION (provided by applicant): Delineating patterns of social affiliation and hierarchical relations in group living rhesus monkeys is very labor intensive. We propose to develop and implement innovative new software and hardware technology that will automate the continuous collection of social interaction data in a complex species-typical rhesus monkey social group. The proposed system will continuously and simultaneously track the precise 3D locations of multiple individuals in a large social group of monkeys in their half acre living area Individual identification will come from comfortable collars, we will design and build, that contai active RFID tags that continuously broadcast each animal's unique code. RFID detectors within and around the compound's periphery allow triangulating each animal's precise location to within 25cm as well as their direction and velocity of movement. We will track the adult males and females that make up the core of the social hierarchy (45 individuals in total). Distance between individuals is a key diagnostic for inferring social affiliation as monkeys are rarely withn 25cm of each other by accident. Thus knowing the amount of time spent at that distance allows inferring affiliative relations. The patterns of affiliation, avoidance, and time in close proximit will allow inferring the social hierarchy of the group. In combination with social network analysis, we will identify patterns of group association and social instability. In addition, we will develop an deploy (in years 3 and 4) a RFID-guided camera system that automatically records high resolution video of predetermined target subjects without human intervention, allowing significantly increased efficiency for later human behavioral coding from these video snippets. Three software systems will be developed under this project: 1) behavioral inference software that infers affiliative, hierarchical, aggressive, and sexual behavior from the automated tracking data; 2) social network analysis software to identify social instabilities; 3) video recognition software that will automate the coding of social behavior from the video snippets collected with the RFID-guided cameras. Together this combined hardware and software will provide an enhanced animal resource making possible easy investigation of the social impact of a variety of manipulations, from drugs to neural insult. In addition, this system can be used to monitor group health and identify individual sickness behavior. This animal resource would facilitate research funded by NIMH, NIDA, or NICHD, and support and enhance primate colony management.