IDBR: Automated Image Analysis for Discovering Motion Behavior in Video

北卡罗莱纳大学夏洛特分校的申博士开发了一种发现生物活动的有趣运动行为的工具，并获得了资助。这通常需要对以复杂模式运动的许多生物实体进行艰苦的观察。这种类型的分析适用于从细胞水平（自然杀伤T细胞，NKT）到种群水平（蜜蜂）的广泛生物系统。然而，行为的复杂性和数据的剪切量使得人工方法不切实际。提出的仪器/工具将全面和自动地在一个大数据集中搜索生物活动的发生，而人工参与最少。该方法包括三个步骤：(1)生物实体运动的自动视频分析，(2)半自动搜索特定的运动行为，(3)使用该工具检验假设。首先，科学家选择一个有趣的运动实例。然后，该工具搜索包含类似动作的视频剪辑。然后，科学家检查并验证搜索结果。如果搜索需要改进，科学家选择好的结果并将其提供给工具。一旦用户对搜索结果有信心，该工具将用于测试与动作相关的特定假设。为了证明该方法在广泛问题上的适用性，将测试涉及蜜蜂通信和自然杀伤T细胞哨兵行为这两种不同生物实体事件的假设。开发一种方法来识别、量化和搜索适用于两种非常不同的生物活动尺度（免疫细胞和蜜蜂）的此类事件的实例的能力，将表明应用于涉及复杂运动行为的各种各样的问题的潜力。拟议的项目将有助于支持本科生的辅导。学生将通过计算机科学系内的REU网站参加REU（本科生研究经验）。生物系有大约1000个本科专业，并继续为本科生研究提供广泛的机会，包括被广泛认为是大学最好的本科荣誉课程之一。此外，北卡罗来纳大学夏洛特分校在北卡罗来纳大学系统的非历史黑人校园中拥有最大的少数民族入学率。这部分是由于大学的城市位置，但也是由于大学的有效性？增加学生群体多样性的美国项目。

A grant has been awarded to Dr. Shin at the University of North Carolina at Charlotte to develop a tool for discovering interesting motion behavior of biological activities. This often requires a painstaking observation of many biological entities moving in complex patterns. Such type of analysis is applicable in a wide range of biological systems from the cellular level (natural killer T cells, NKT) to the population level (honeybees). However, the complexity of the behavior and the shear amount of data makes a manual approach impractical. The proposed instrument/tool will comprehensively and automatically search for the occurrences of a biological activity in a large dataset with minimal manual involvement.The method involves three steps: (1) automated video analysis of biological entities' motion, (2) semi-automated searching for a specific motion behavior, and (3) testing the hypothesis using the tool. First, a scientist selects one instance of interesting motion. Then, the tool searches for video clips that contain similar motion. The scientist then examines and validates the search results. If the search needs improvement, the scientist selects the good results and provides them to the tool. Once the user is confident with the search results, the tool will be used for testing a specific hypothesis related to the motion. To demonstrate the applicability of the method on a wide range of problems, hypotheses involving events of two different biological entities of honey bee communication and natural killer T cell sentry behavior will be tested. The ability to develop an approach to identifying, quantifying and then searching for instances of such events that is applicable to two very different scales of biological activity (immune cells and honeybees) would suggest the potential for application to a wide variety of problems involving complex motion behaviors.The proposed project will help support mentoring of undergraduate students. The students will participate in the REU (research experience for undergraduates) through the REU site within the Department of Computer Science. The Biology Department has approximately 1000 undergraduate majors and continues to offer extensive opportunities for undergraduate research including an undergraduate honors program widely considered to be one of the best in the University. In addition, UNC Charlotte holds the distinction of the largest minority enrollment of any of the non-historically black campuses of the UNC system. This is partly due to the urban location of the University, but also to the effectiveness of the University?s programs for increasing the diversity of student body.