• PKU-LF
• • • Abstract
    • Quick Comparison
    • Dataset
    • Citation

PKU-LF

A large-scale light field dataset developed for multiple research fields. If you find that our data violates your rights, please contact us immediately, and we will deal with it in time.

Abstract

Compared with current RGB or RGB-D saliency detection datasets, those for light field saliency detection often suffer from many defects, e.g., insufficient data amount and diversity, incomplete data formats, and rough annotations, thus impeding the prosperity of this field. To settle these issues, we elaborately build a large-scale light field dataset, dubbed PKU-LF, comprising 5,000 light fields and covering diverse indoor and outdoor scenes. Our PKU-LF provides all-inclusive representation formats of light fields and offers a unified platform for comparing algorithms utilizing different input formats. For sparking new vitality in saliency detection tasks, we present many unexplored scenarios (such as underwater and high-resolution scenes) and the richest annotations (such as scribble annotations, bounding boxes, object-/instance-level annotations, and edge annotations), on which many potential attention modeling tasks can be investigated. To facilitate the development of saliency detection, we systematically evaluate and analyze 16 representative 2D, 3D, and 4D methods on four existing datasets and the proposed dataset, furnishing a thorough benchmark. Furthermore, tailored to the distinct structural characteristics of light fields, a novel symmetric two-stream architecture (STSA) network is proposed to predict the saliency of light fields more accurately. Specifically, our STSA incorporates a focalness interweavement module (FIM) and three partial decoder modules (PDM). The former is designed to efficiently establish long-range dependencies across focal slices, while the latter aims to effectively aggregate the extracted coadjutant features in a mutual-enhancement way. Extensive experiments demonstrate that our method can significantly outperform the competitors.

Figure 1: Illustration of light fields. (a) Theory of light fields. (b) Light field data in our PKU-LF.

Quick Comparison

We provide the results in our paper for a quick comparison, and you can download the predicted saliency maps of our method on different datasets from here.

Dataset

Usage of our dataset is under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International Public License.

Figure 2: Rich annotations of the proposed PKU-LF dataset including scribble annotations, bounding boxes, object-/instance-level annotations, and edge annotations.

Dataset	Calibration data	Raw data	All-focus image	Depth map	Focal stack	Relative depth of field	Sub-aperture image & Micro-lens image	Annotation
PKU-LF	Download	Download	Download	Download	Download	Download	Download	Download

*Note: We made two major upgrades to the dataset. First, some annotation errors were corrected. Second, we expanded the number of samples to 9909 to meet the needs of existing learning paradigms for large-scale data.

Citation

Please cite our paper if you find our work is helpful.

@article{gao2023thorough,
  title={A Thorough Benchmark and a New Model for Light Field Saliency Detection},
  author={Gao, Wei and Fan, Songlin and Li, Ge and Lin, Weisi},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2023},
  publisher={IEEE}
}