In this paper, we address the generation of semantic labels describing the headgear\naccessories carried out by people in a scene under surveillance, only using depth information\nobtained from a Time-of-Flight (ToF) camera placed in an overhead position. We propose a new\nmethod for headgear accessories classification based on the design of a robust processing strategy that\nincludes the estimation of a meaningful feature vector that provides the relevant information about\nthe people�s head and shoulder areas. This paper includes a detailed description of the proposed\nalgorithmic approach, and the results obtained in tests with persons with and without headgear\naccessories, and with different types of hats and caps. In order to evaluate the proposal, a wide\nexperimental validation has been carried out on a fully labeled database (that has been made available\nto the scientific community), including a broad variety of people and headgear accessories. For the\nvalidation, three different levels of detail have been defined, considering a different number of classes:\nthe first level only includes two classes (hat/cap, and no hat/cap), the second one considers three\nclasses (hat, cap and no hat/cap), and the last one includes the full class set with the five classes\n(no hat/cap, cap, small size hat, medium size hat, and large size hat). The achieved performance is\nsatisfactory in every case: the average classification rates for the first level reaches 95.25%, for the\nsecond one is 92.34%, and for the full class set equals 84.60%. In addition, the online stage processing\ntime is 5.75 ms per frame in a standard PC, thus allowing for real-time operation.
Loading....