Wifbs: A web-based image feature benchmark system

Abstract

Automatic analysis of image data is of high importance for many applications. Given an image classification problem one needs three things: (i) Training data and tools to extract (ii) relevant visual information— usually image features—that can be used by (iii) classification algorithms. For given (i), a multitude of candidates present themselves for (ii) and (iii). Model selection becomes the main issue. We present a webbased feature benchmark system enabling system designers to streamline tool-chains to specific needs using available implementations of candidate tools. Our system features a modular architecture, remote and parallel computing, extensibility and—from a user’s standpoint—platform independence due to its web-based nature. Using Wifbs1, image features can be subjected to a sophisticated and unbiased model selection procedure to compose optimized pipelines for given image classification problems.