Google-Landmarks 谷歌地标数据集

#### 背景描述
你有没有浏览过你的假期照片并问自己：我在中国参观的这座寺庙叫什么名字？谁创造了我在法国看到的这座纪念碑？地标识别可以提供帮助！该技术可以直接从图像像素预测地标标签，以帮助人们更好地理解和组织他们的照片集。今天，地标识别研究的一大障碍是缺乏大型注释数据集。这促使我们发布了迄今为止最大的全球数据集 Google-Landmarks，以促进在这个问题上取得进展。
#### 数据说明
数据集分为两组图像，以评估两种不同的计算机视觉任务：识别和检索。该数据最初在 [1] 中进行了描述，并作为 Google Landmark Recognition Challenge 和 Google Landmark Retrieval Challenge 的一部分发布。
此外，为了促进该领域的研究，我们开源了深度局部特征 (DELF)，这是一种我们认为特别适合此类任务的周到的局部特征描述符。 DELF 的代码可以通过这个链接在 github 上找到。
#### 数据来源
H. Noh, A. Araujo, J. Sim, T. Weyand, B. Han, "Large-Scale Image Retrieval with Attentive Deep Local Features", Proc. ICCV'17

If you make use of the Google Landmark Boxes dataset in your research, please consider citing:

M. Teichmann*, A. Araujo*, M. Zhu and J. Sim, “Detect-to-Retrieve: Efficient Regional Aggregation for Image Search”, Proc. CVPR'19
#### 问题描述
计算机视觉

#### Background
Have you ever browsed your vacation photos and wondered: What is the name of this temple I visited in China? Who created this monument I saw in France? Landmark recognition can help! This technology can directly predict landmark labels from image pixels to help people better understand and organize their photo collections. Today, a major barrier to landmark recognition research is the lack of large-scale annotated datasets. This prompted us to release Google-Landmarks, the largest global dataset to date, to advance progress on this problem.
#### Data Description
The dataset is split into two image sets to evaluate two distinct computer vision tasks: recognition and retrieval. This data was originally described in [1] and released as part of the Google Landmark Recognition Challenge and Google Landmark Retrieval Challenge.
Additionally, to promote research in this domain, we have open-sourced Deep Local Feature (DELF), a well-designed local feature descriptor that we deem particularly suitable for such tasks. The code for DELF can be found on GitHub via the provided link.
#### Data Source
H. Noh, A. Araujo, J. Sim, T. Weyand, B. Han, "Large-Scale Image Retrieval with Attentive Deep Local Features", Proc. ICCV'17

If you make use of the Google Landmark Boxes dataset in your research, please consider citing:

M. Teichmann*, A. Araujo*, M. Zhu and J. Sim, "Detect-to-Retrieve: Efficient Regional Aggregation for Image Search", Proc. CVPR'19
#### Problem Description
Computer Vision