视觉词袋：什么是合理的词（矢量）维度？ [英] Bag of Visual Words: what is a reasonable word (vector) dimension?

问题描述

在一揽子特征/视觉词汇范例中，我们有一个载体 V in k -dimensions，其中 V [i] = j 如果 i -th centroid（由 k -means算法获得）是所有<$中最接近的一个c $ c> k -centroids for j 视觉描述符（例如SIFT描述符）。

In the Bag of Features/Visual Words paradigm we have a vector V in k-dimensions, where V[i]=j if the i-th centroid (obtained by k-means algorithm) is the closest one among all the k-centroids for j visual descriptors (e.g. SIFT descriptors).

AFAIK，由此产生的视觉向量非常稀疏（这意味着大多数条目都是0值）因为 k 非常大，但我的问题是：数据是 k 的合理价值（以及矢量大小）？数百个维度？成千上万的？特别是考虑到 k -means执行时间取决于 k 。

AFAIK, the resulting visual vector is very sparse (it means that most of entries are 0-value) since k is really big, but my question is: what is a reasonable value for k (and so the vector size)? Hundreds of dimensions? Thousands? Especially considering that k-means execution time depends from k.

推荐答案

真的，取决于你的数据。以下是经验法则：

Depends on your data, really. Here is the rule of thumb:

太小K：您的群集不代表所有补丁。
太大K：你可能会得到量化伪像并且可能过度拟合。

Too small K: your clusters will not represent for all patches. Too large K: you may get quantization artifacts and probably overfitting.

这篇关于视觉词袋：什么是合理的词（矢量）维度？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！