This paper describes progress in developing a computer vision system for the automated detection of pedestrians. It describes the results of recent trials of the pedestrian detector, during which the accuracy of pedestrian presence detection and volume estimation were evaluated. The development of the PUFFIN (Pedestrian Use FFriendly INtelligent) pedestrian crossing is especially important. A key factor in the development of vision algorithms is the quality of the evaluation system used to assess their performance. The most important characteristics of the evaluation system are that it is representative, repeatable, and, as far as possible, automated. The paper describes an evaluation system to meet these requirements, based on digitisation and one-off manual analysis of test video sequences. An evaluation data set was developed, to ensure that test data represent the range of conditions that the system is likely to meet during its operation; many different factors need to be considered here. Results are given of the assessment of pedestrian algorithms using these methods; they include both binary detection results and volumetric detection results. The repeatability of evaluation has led to a better understanding of the algorithms. Reliance on manual input has been reduced greatly, but full automation is not yet achieved.
Abstract