tailieunhanh - Multi-agent event recognition in structured scenarios

Even moderate-sized events take a lot of planning and organising. It’s wise to set up an organising group right from the start and allocate areas of responsibility. Extra helpers can be brought in for particular roles, such as stewarding, without having to attend all the planning meetings. This approach cuts down on stress for particular individuals and ensures a more efficient use of everyone’s time and energy. For more about communications with your entire team see the section on organising your team. . | Multi-agent event recognition in structured scenarios Vlad I. Morariu and Larry S. Davis Institute for Advanced Computer Studies University of Maryland College Park MD 20742 morariu lsd @ Abstract We present a framework for the automatic recognition of complex multi-agent events in settings where structure is imposed by rules that agents must follow while performing activities. Given semantic spatio-temporal descriptions ofwhat generally happens . rules event descriptions physical constraints and based on video analysis we determine the events that occurred. Knowledge about spatiotemporal structure is encoded using first-order logic using an approach based on Allen s Interval Logic and robustness to low-level observation uncertainty is provided by Markov Logic Networks MLN . Our main contribution is that we integrate interval-based temporal reasoning with probabilistic logical inference relying on an efficient bottom-up grounding scheme to avoid combinatorial explosion. Applied to one-on-one basketball our framework detects and tracks players their hands and feet and the ball generates event observations from the resulting trajectories and performs probabilistic logical inference to determine the most consistent sequence of events. We demonstrate our approach on 1hr 100 000frames of outdoor videos. 1. Introduction The automated analysis of multi-agent activity is difficult due to interactions that lead to large state spaces and complicate the already uncertain low-level processing. Often activities must satisfy rules that impose a spatio-temporal structure on constituent events. This structure can be leveraged to disambiguate amongst complex activities. For example in the case of one-on-one basketball offensive and defensive rebounds are often ambiguous since both players are near each other as they reach for the ball. However the rules of half-court basketball can reduce this ambiguity by relating the rebound event to other less ambiguous events .