[CV] Retrieval

Introduction

Retrieval (图搜) 也是CV领域一个应用非常广泛的方向,在安防场景下,我们常常会根据摄像头抓怕的嫌疑人头像去人脸数据库里进行搜索;在电商平台,用户也会拍摄图片进行上传,我们的算法应返回对应的SKU,此为“拍照购”。Retrieval的难点主要在于如何训练表示能力良好的embedding模型,以及如何高效地进行feature similarity searching,尤其是库里SKU种类达到亿级别的时候,为了保证良好的用户体验,如何能快速而准确地给用户返回匹配信息,是一直以来不少研究者和工程师致力于解决的问题。由于Deep Learning的飞速发展,Retrieval在工业应用方面越来越成熟,常规流程如下:

  1. 训练embedding模型,可视为一个Classification或Metric Learning问题
  2. 如果对检索速度有高要求,可能需要做Hash或Quantization
  3. 利用FAISS做大规模相似性搜索
  4. Similarity衡量标准一般是$L_2$ distance或Cosine Similarity

本文主要分享一些读过的顶会/顶刊上的paper。

@LucasXU注:本文长期更新。

Supervised Deep Hashing for Scalable Face Image Retrieval

Paper: Supervised Deep Hashing for Scalable Face Image Retrieval

这是一篇Face Retrieval方向的文章,整体framework和idea也非常简单:

  1. Deep hash的引入,Multi-task Loss: 同时优化Classification Loss和Quantization Loss。
  2. low-level和high-level information的fusion,来获取multi-scale的信息

因Retrieval场景的特殊性(亿级别的item + 高维特征向量匹配),以及用户对速度与精度的需求,不少Retrieval方法会采用Hashing来生成图像的compact binary codes,而binary codes的Similarity Search会非常快:

  1. Hamming distance的计算可以仅通过XOR operation得到。
  2. 将highly compressed data加载进内存,减小了大容量内存的需求。

当前的hashing方法主要有两种:

  1. Data-independent: 使用random projection来产生binary codes,例如Locality-Sensitive Hashing (LSH)。
  2. Data-dependent: 在尽可能保留data structure的情况下从数据中学习Hashing function。

而Learning-based Hashing methods又可以被分为unsupervised hashing (例如random projection, reconstruction error minimization, graph-based hashing, quantization error minimization) 和supervised hashing两类。

本文模型网络结构图如下:
DHCQ

Learning-based hashing methods旨在学习某种hashing function来为每张图生成compact binary codes,即$X\to B\in \{0,1\}^{k\times N}$,$k$为binary codes的length。

在实验中,作者用了一种非常Naive的方法来进行quantize,即将最后一个隐层的输出作为sigmoid function的输入,使其被squeeze到$(0, 1)$区间,然后再通过符号函数二值化:
$$
sign(x)=\begin{cases}
1 & if x\geq 0.5 \\
0 & otherwise
\end{cases}
$$

Softmax Loss作为classification loss,$L_2$ loss作为quantization loss:
$$
min |B-H|_F^2
$$
其中,$B=sign(H)$,通过优化quantization loss,$H$会越来越接近1或者0。

实验中,作者发现,classification criterion比quantization loss在explore discriminative information方面更重要。

Reference

  1. Zhao, Bo, et al. “Memory-augmented attribute manipulation networks for interactive fashion search.” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017.
  2. Chum, Ondrej, et al. “Total recall: Automatic query expansion with a generative feature model for object retrieval.” 2007 IEEE 11th International Conference on Computer Vision. IEEE, 2007.
  3. Tang, Jinhui, Zechao Li, and Xiang Zhu. “Supervised deep hashing for scalable face image retrieval.” Pattern Recognition 75 (2018): 25-32.
  4. Radenović, Filip, Giorgos Tolias, and Ondrej Chum. “Fine-tuning CNN image retrieval with no human annotation.” IEEE transactions on pattern analysis and machine intelligence (2018).
  5. Babenko, Artem, and Victor Lempitsky. “Aggregating local deep features for image retrieval.” Proceedings of the IEEE international conference on computer vision. 2015.
  6. Mousavian, Arsalan, and Jana Kosecka. “Deep convolutional features for image based retrieval and scene categorization.” arXiv preprint arXiv:1509.06033 (2015).
  7. Gordo, Albert, et al. “Deep image retrieval: Learning global representations for image search.” European conference on computer vision. Springer, Cham, 2016.
  8. Huang, Junshi, et al. “Cross-domain image retrieval with a dual attribute-aware ranking network.” Proceedings of the IEEE international conference on computer vision. 2015.
  9. Xie, Lingxi, et al. “Image classification and retrieval are one.” Proceedings of the 5th ACM on International Conference on Multimedia Retrieval. Acm, 2015.

Powered by Hexo and Hexo-theme-hiker

Copyright © 2018 - 2024 LucasXU All Rights Reserved.

UV : | PV :