Introduction to the

CVPR stands for IEEE Conference on Computer Vision and Pattern Recognition. The conference is the premier conference in the field of computer vision and pattern recognition organized by IEEE.

It is the annual IEEE conference on computer vision and pattern recognition technology. CVPR is one of the world’s top computer vision conferences (ICCV and ECCV are the other two). CVPR has a fixed theme every year, and every year, companies sponsor the conference and get the opportunity to present at the conference.

CVPR 2019 saw more than 5,100 submissions, of which 1,300 were accepted, achieving an acceptance rate of nearly 25.2%.

CVPR 2019 Accepted papers have been released and can be viewed on the official website:

openaccess.thecvf.com/CVPR2019.py

As follows:

Pay attention to my public account — the growth of algorithmic ape, background reply “CVPR2019”, get the network disk link of all papers.

In addition, in zhuanlan.zhihu.com/p/68679900 below to see this article share a direct crawl website to download code:

Github.com/Simon32/CVP…

After a simple test, this code is available, of course, the speed may not be very fast, after all, just a simple implementation of the paper download function. Also, all the papers together need about 2 gigabytes of storage space.

Visual website

After a brief introduction of CVPR2019, we can know that a total of 1300 papers have been accepted, but how to quickly find the papers you need from these papers? Although the official website does provide a search bar, which can be searched according to keywords, the results are not accurate enough.

Here are two great references, the latter of which I have previously recommended: [Github Project] a better place to read and find papers

  • Github.com/karpathy/ni…
  • Github.com/karpathy/ar…

I have made a visual website about CVPR 2019. To be precise, I can quickly check the abstract and keywords of the paper before judging whether it is worth further browsing or perusing. The link on the official website is as follows:

mattdeitke.github.io/CVPR-2019/

As shown below:

The features of this site are:

  1. Rank papers according to LDA topic model;
  2. The content of each paper can be selected, including abstract, keywords, bibtex, and abbreviated PDF

To be specific:

Firstly, it uses LDA(Latent Dirichlet Allocation) algorithm to cluster all papers, and then classify them according to different topics. K =7 is set here, namely seven categories, which are as follows:

  • 0 means video
  • 1 is for geometry
  • 2 represents image processing
  • 3 indicates neural network clipping
  • 4 indicates graphic description
  • 5 means split
  • 6 means unsupervised learning

Here you can choose which topic to sort the papers by, and then you can choose what each paper presents. Both the abstract and the ABBREVIATED PDF are good, which is also available at Arxiv Sanity. What is new is that the Top 100 words that appear most frequently in each paper are based on the LDA model. According to these words, you can get a general idea of the content of the paper.

Finally, this big god also open source this code, according to the big god, as long as the previous CVPR webpage does not change, can also be used, that is, not limited to CVPR2019, CVPR2018, 2017, etc., of course, the future CVPR2020 can also be used.

Github.com/mattdeitke/…

Specific installation and use methods are shown in the figure below:

Finally, follow my public account — the growth of algorithmic ape, and reply “CVPR2019” in the background to obtain the network disk link of all papers.

CVPR2019 Accepted papers’ official website:

openaccess.thecvf.com/CVPR2019.py

Visual official website:

mattdeitke.github.io/CVPR-2019/

Welcome to follow my wechat official account — the growth of algorithmic ape, or scan the QR code below, we can communicate, learn and progress together!