In film art, the usage of various fields of view in a movie reflects the shooting style of the director or artist. Analyzing which field of view was used by the shot can help us achieve the semi-automatic design and generation of the storyboard. Currently,although there are a growing number of image datasets for aesthetic metrics, text-to-image generation and emotional metrics, datasets of cinematographic footage exclusively are extremely rare. So, we build a dataset named Movie Field of View Dataset (MFOVD1.0) including 21541 images. Our dataset covers excellent films of different eras at home and abroad, as well as winning films of well-known film awards such as the Oscars. Our work is to capture footage from the original films of these films, and classify these shots into five categories: long shot, full shot, medium shot, close and big close-up.
If you want to further understand and use MFOVD1.0, please apply by Request access.doc and send it to jinxinbesti@foxmail.com. We will tell you how to have an access to MFOVD1.0 by email according to your application.
Xin Jin, Chenyu Fan, Biao Wang, Yihang Bo*, Xinzhe Pan, Zihan Jia, Ya Zhuo, Runqi Zhang, Shuai Cui. FOV Recognizer: Telling the Field of View of Movie Shots. The 5th Chinese Conference on Pattern Recognition and Computer Vision (PRCV), Shenzhen, China, 14-17 October, 2022.
Please cite the PRCV paper if you use MFOVD1.0 in your work:
@inproceedings{
author = {Xin Jin, Chenyu Fan, Biao Wang, Yihang Bo*, Xinzhe Pan, Zihan Jia, Ya Zhuo, Runqi Zhang, Shuai Cui},
title = {FOV Recognizer: Telling the Field of View of Movie Shots},
booktitle = {The 5th Chinese Conference on Pattern Recognition and Computer Vision (PRCV), Shenzhen, China, 14-17 October, 2022},
year = {2022},
}