Multimedia Retrieval Benchmark Suite And Characteristics Analysis

Multimedia data, especially image and video data, have become one of the most overwhelming data types on the Internet recently. Considering the user experience and real application requirements, multimedia data always demand a real-time processing speed. As a result, the huge amounts of such data make retrieving useful information from them not only data-intensive, but also computation-intensive, which poses significant challenges to current system and architecture designs. Unfortunately, most prior studies focus only on text-based retrieval systems or traditional multimedia processing applications. As far as we know, there is no systematic study on analyzing the characteristics of multimedia retrieval applications and how they might impact system and architecture designs.In this paper, we make the first attempt to construct a multimedia retrieval benchmark suite to evaluate the corresponding system and architecture designs. To embody diverse multimedia retrieval applications, we collect eight state-of-art multimedia retrieval algorithms which cover the whole retrieval stages, including feature extraction, feature matching, and spatial verification. Based on the benchmark constructed, we further analyze the inherent architecture characteristics, which provide some insights into system and architecture design for multimedia retrieval applications. In summary, this paper makes the following contributions:●Design and implementation of a multimedia retrieval benchmark suite:(?) Include a set of state-of-art multimedia retrieval algorithms(?) Cover the whole retrieval stages, including feature extraction, feature matching and spatial verification(?) Provide input sets in different sizes to enable the algorithms to be used both individually and combined together for system-level evaluation(?) Provide a set of flexible interfaces to assign parameters automatically, and design a basic framework to construct an image retrieval system easily.(?) Architectural characteristic analysis for multimedia retrieval algorithms, which provides some insights to architecture design and system evaluation for these algorithms:(?) Complex branch behavior with higher branch misprediction rate(?) Lower potential instruction-level parallelism (ILP)(?) Poorer data temporal locality and better data spatial locality under certain circumstance(?) Floating-operation insensitive(?) Memory and computation intensive(?) Imbalance workload and high bandwidth requirement among different retrieval stages and very large-scale data in the backend database...