AI Paper Atlas

Research map / Benchmarks / BSDS500

BSDS500 leaderboard

15 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on BSDS500. Best ODS: 0.848 (EDTER-MS-VOC, CVPR 2022).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

BSDS500 · ODS

14 papers, each with its best reported number for its own method; higher is better.

#ODS ↑MethodPaperTopic
10.848EDTER-MS-VOCEDTER: Edge Detection With Transformer
CVPR 2022 · Table: Results on BSDS500 test
CNN features
20.839SAUGE with SOC and guide loss
SAM ViT-B
SAUGE: Taming SAM for Uncertainty-Aligned Multi-Granularity Edge Detection
AAAI 2025 · Table 5
Open-set and few-shot detection
30.828BDCN ‡
VGG16
Bi-Directional Cascade Network for Perceptual Edge Detection
CVPR 2019 · Table: Comparison with other m
CNN features
40.815CED-VOC-aug
HED
Deep Crisp Boundaries
CVPR 2017 · Table 2
CNN features
50.805PGA-Net (fusion) w/ CK (5S)
ResNet50
Probabilistic Graph Attention Network with Conditional Kernels for Pixel-Wise Prediction
TPAMI 2020 · Table 1
CNN features
60.798OD-SRN
SRN
Orthogonal Decomposition Network for Pixel-Wise Binary Classification
CVPR 2019 · Table 5
CNN features
70.798AMH-Net (fusion)
ResNet50
Learning Deep Structured Multi-Scale Features using Attention-Gated CRFs for Contour Prediction
NeurIPS 2017 · Table: BSDS500 dataset: quanti
CNN design
80.788CEDN
CEDN pretrained on PASCAL VOC
Object Contour Detection With a Fully Convolutional Encoder-Decoder Network
CVPR 2016 · Table: Contour detection resul
CNN features
90.788BNF
DeepLab FCN
Semantic Segmentation With Boundary Neural Fields
CVPR 2016 · Table 1
CNN features
100.782HED (ours)Holistically-Nested Edge Detection
IJCV 2017 · Table (BSDS500)
CNN features
4 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

BSDS500 · AP

10 papers, each with its best reported number for its own method; higher is better.

#AP ↑MethodPaperTopic
10.903EDTER-MS-VOCEDTER: Edge Detection With Transformer
CVPR 2022 · Table: Results on BSDS500 test
CNN features
20.889CED-VOC-aug
HED
Deep Crisp Boundaries
CVPR 2017 · Table 2
CNN features
30.888BDCN (+PASCAL-Context)
VGG16
BDCN: Bi-Directional Cascade Network for Perceptual Edge Detection
TPAMI 2020 · Table BSDS500 test
CNN features
40.869AMH-Net (fusion)
ResNet50
Learning Deep Structured Multi-Scale Features using Attention-Gated CRFs for Contour Prediction
NeurIPS 2017 · Table: BSDS500 dataset: quanti
CNN design
50.851BNF
DeepLab FCN
Semantic Segmentation With Boundary Neural Fields
CVPR 2016 · Table 1
CNN features
60.833HED (ours)Holistically-Nested Edge Detection
IJCV 2017 · Table (BSDS500)
CNN features
70.821CEDN
CEDN pretrained on PASCAL VOC
Object Contour Detection With a Fully Convolutional Encoder-Decoder Network
CVPR 2016 · Table: Contour detection resul
CNN features
80.82Ours (patch, K = 16 bits)Unsupervised Learning of Discriminative Attributes and Visual Representations
CVPR 2016 · Table 4
Open-set and few-shot detection
90.782OD-SRN
SRN
Orthogonal Decomposition Network for Pixel-Wise Binary Classification
CVPR 2019 · Table 5
CNN features
100.78SemiContour
Structured random forests
SemiContour: A Semi-Supervised Learning Approach for Contour Detection
CVPR 2016 · Table 1
Semi-supervised learning

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.