Web-Scale Responsive Visual Search at Bing.

In this paper, we introduce a web-scale general visual search system deployedin Microsoft Bing. The system accommodates tens of billions of images in theindex, with thousands of features for each image, and can respond in less than200 ms. In order to overcome the challenges in relevance, latency, andscalability in such large scale of data, we employ a cascaded learning-to-rankframework based on various latest deep learning visual features, and deploy ina distributed heterogeneous computing platform. Quantitative and qualitativeexperiments show that our system is able to support various applications onBing website and apps.