Evaluation of video detection systems, volume 1 : effects of configuration changes in the performance of video detection systems.

The effects of modifying the configuration of three video detection (VD) systems (Iteris, Autoscope, and Peek) are evaluated in daytime and nighttime conditions. Four types of errors were used: false, missed, stuck-on, and dropped calls. The three VD systems were installed side-by-side at an intersection in Rantoul, IL. The configurations were modified by the vendors to improve their performance. The modifications to Peek VD configuration effectively reduced dropped calls at the stop bar zones; however, that was at the expense of increasing false calls during daytime and missed and false calls during night time. Similarly, in the advance zones, in both daytime and nighttime, there was a clear tradeoff between decreasing missed calls and increasing false calls. The modifications to Autoscope VD configuration did not provide a clear improvement at the stop bar zones during daytime; however, during nighttime, false calls increased and missed calls were eliminated. In the advance zones, the Autoscope changes significantly reduced missed calls in both day and night, reduced false calls in daytime, but increased in false calls during nighttime. The modifications to Iteris VD configuration were slight and overall effects of the changes were relatively small. This resulted in a tradeoff between false and missed calls. The results for three systems indicate that there are tradeoffs when the goal is to improve the overall performance of VD systems. Thus, after making modifications to the configuration of VD systems, the effects of these changes should be monitored not only for improvements on the previously detected errors, but also for potential new errors of a different type.