Google Goggles 用的算法是怎么样的

参考这里:https://wiki.engr.illinois.edu/download/attachments/197310410/google+goggles.pdf?version=1\u0026amp;modificationDate=1304456114000还可参考这里:http://www.quora.com/Algorithms/What-is-the-algorithm-used-by-Google-Search-by-Image-1
■网友
图图搜是先找到淘宝上的同款,然后拿到产品tag,接着根据tag、主颜色等信息进行二次查找。最基本的技术还是相同图像查找,当然也包含了商品主体识别。但是对于真实自然场景下的服装图像效果就不是很好了。
■网友
For finding more images of the exact same physical object (e.g., a Pepsi can) or scene (e.g.,another photo of Eiffel Tower under similar illumination), researchers typically use the general bag-of-words paradigm introduced by the Video Google work , where a large histogram of quantized local image patches (usually encoded with the SIFT descriptor) is used for image retrieval. This paradigm generally works extremely well (especially for heavily-textured objects),and has led to many successful applications such as GOOGLE GOGGLES.However, these methods usually fail when tasked with finding similar, but not identical objects (e.g., try using GOOGLE GOGGLES app to find a cup, or a chair). This is because SIFT, being a local descriptor, captures the minute details of a particular object well, but not its overall global properties


    推荐阅读