Pets vs. vace evaluation programs A comparative study.pdfVIP

PETS vs. VACE Evaluation Programs: A Comparative Study Vasant Manohar, Matthew Boonstra, Valentina Korzhova, Padmanabhan Soundararajan, Dmitry Goldgof and Rangachar Kasturi Computer Science & Engineering, University of South Florida, Tampa, FL {vmanohar, boonstra, korzhova, psoundar, goldgof, r1k}@ Shubha Prasad and Harish Raju Video Mining Inc., State College, PA {sprasad, hraju}@ Rachel Bowers and John Garofolo National Institute of Standards and Technology, Gaithersburg, MD {rachel.bowers, john.garofolo}@ Abstract There are several current efforts towards the evaluation of object detection and tracking in video. The Video Anal- There are numerous distributed efforts on performance ysis and Content Extraction (VACE) program, supported by evaluation of computer vision algorithms. Information ex- Advanced Research and Development Activity (ARDA) is change between programs can expedite achievement of the one such endeavor, established with the objective of devel- common goals of such projects. As a step towards this, this oping novel algorithms and implementations for automatic paper presents a qualitative comparison of the VACE and video content extraction, multi-modal fusion and event un- the PETS programs, the synergy of which, we believe, will derstanding. The program has several phases and is cur- have a signi?cant impact on the progress of research in the rently nearing the end of Phase II. During VACE Phase I and vision community. The comparison is based on the vital as- Phase II, the program achieved signi?cant progress in vid