Managing High Availability (HA) among Computer Vision Systems deployed on Edge devices requires expensive hardware redundancy. The cost of scaling such systems becomes prohibitive. We present an alternative approach for managing high availability by building this capacity at an application layer. To this end, we propose the use of pattern recognition components to make the application context aware. This intelligent application is then itself able to decide which Edge modules are running redundant or less significant processes. For example, if the module is processing uneventful video streams, the application may decide to make this compute available to some other module. In the event of hardware failures, this application can then provision workloads on idle hardware (or hardware running less significant workload).