A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics

By David R. Martin, Charless C. Fowlkes, Doron Tal, Jitendra Malik. 2001

8,885 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

In Evaluation & Benchmarks

Citations in the corpus

Neural Autoregressive Distribution Estimation (2016, methodology): “Following the work of Zoran and Weiss (2011), we use 8-by-8-pixel patches of monochrome natural images, obtained from the BSDS300 dataset (Martin et al., 2001; Figure 9 gives examples).”
Semantic Understanding of Scenes Through the ADE20K Dataset (2016, background): “ssed some small objects in different annotations. Missing labels account for 5.9% of the error pixels. A similar issue existed in segmentation datasets such as the Berkeley Image segmentation dataset [21]. The median error values for the three error types are: 4.8%, 0.3% and 2.6% showing that the me”
Deep Watershed Transform for Instance Segmentation (2016, background): “, Pascal VOC [8], BSDS500 [17], and CVPPP [19]) Cityscapes has a large number of scenes involving dozens of instances with large degrees of occlusions at vastly different scales.”
Benchmarking Deep Reinforcement Learning for Continuous Control (2016, neutral): “, 2010), BSDS500 (Martin et al., 2001), SWITCHBOARD (Godfrey et al.”
SUN database: Large-scale scene recognition from abbey to zoo (2010, methodology): “Texton Histograms: We build a 512 entry universal texton dictionary [18] by clustering responses to a bank of filters with 8 orientations, 2 scales, and 2 elongations.”
LVIS: A Dataset for Large Vocabulary Instance Segmentation (2019, methodology): “We owe much of our current success in image recognition to pioneering datasets such as MNIST [16], BSDS [20], Caltech 101 [6], PASCAL VOC [5], ImageNet [23], and Toy Deer”
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (2016, methodology): “Mean opinion score (MOS) testing In all conducted MOS tests we have asked 26 human raters to assign a score from 1 (Bad) to 5 (Excellent) to reconstructions of the 4× downsampled versions of images from Set5, Set14 and BSD100.”
Segment Anything (2023, methodology): “We evaluate SAM on the classic low-level task of edge detection using BSDS500 [72, 3].”
Fast Edge Detection Using Structured Forests (2014, methodology): “The majority of our experiments are performed on the Berkeley Segmentation Dataset and Benchmark (BSDS500) [1], [35].”
CIDEr: Consensus-based image description evaluation (2014, background): “This has already been shown in various problems in computer vision, such as detection [13, 7], segmentation [13, 28], and stereo [39].”
Scene Parsing through ADE20K Dataset (2017, background): “A similar issue existed in segmentation datasets such as the Berkeley Image segmentation dataset [20].”
Generative Image Modeling Using Spatial LSTMs (2015, methodology): “Although our model’s assumed strength lies in its ability to scale to large images and to capture long-range correlations, we include results on BSDS300 to make a connection to this part of the literature.”
UberNet: Training a Universal Convolutional Neural Network for Low-, Mid-, and High-Level Vision Using Diverse Datasets and Limited Memory (2016, methodology): “3 high-level annotation (e.g. object positions, landmarks in PASCAL VOC [26]) is often missing from the datasets used for low-level tasks (e.g. BSD [69]), and vice versa.”
Masked Autoregressive Flow for Density Estimation (2017, methodology): “This dataset was obtained by extracting random 8×8 monochrome patches from the BSDS300 dataset of natural images [20].”
A Comprehensive guide to Bayesian Convolutional Neural Network with Variational Inference (2019, methodology): “The Network architecture was trained on BSD300 dataset [46] provided by the Berkeley Computer Vision Department.”
AutoGAN-Distiller: Searching to Compress Generative Adversarial Networks (2020, neutral): “We evaluate the searched/compressed model on several popular Algorithm 1 The Proposed AutoGAN-Distiller Framework 1: Input: dataset χ = { x i } Ni =1 , pretrained generator G 0 , search space and supernet G , epochs to pretrain ( T 1 ), search ( T 2 ) and train-from-scratch ( T 3 ) 2: Output: traine”