AI Paper Atlas

Research map / Benchmarks / Pascal VOC

Pascal VOC leaderboard

73 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on Pascal VOC. Best mIoU: 97.2 (SCAN, CVPR 2024).

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.

Pascal VOC · mIoU

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

#mIoU ↑MethodPaperTopic
197.2SCAN
CLIP ViT-L/14
Open-Vocabulary Segmentation with Semantic-Assisted Calibration
CVPR 2024 · Table 1
Open-set and few-shot detection
294.6SAN
CLIP ViT-L/14
Side Adapter Network for Open-Vocabulary Semantic Segmentation
CVPR 2023 · Table: Performance comparison
Open-set and few-shot detection
388.5FLANet
ResNet150
Fully Attentional Network for Semantic Segmentation
AAAI 2022 · Table 6
CNN features
488.2EMANet
ResNet-152
Expectation-Maximization Attention Networks for Semantic Segmentation
ICCV 2019 · Table 2
CNN features
588.1Our proposed (Xception-65)
Xception-65
Decoders Matter for Semantic Segmentation: Data-Dependent Decoding Enables Flexible Feature Aggregation
CVPR 2019 · Table: State-of-the-art method
CNN features
688DeepLabv3+ with Euclidean Frank-Wolfe CRF
DeepLabv3+
Regularized Frank-Wolfe for Dense CRFs: Generalizing Mean Field and Beyond
NeurIPS 2021 · Table: Performance on the test
Kernels and regression theory
788HEMNet
ResNet-101
Improving Gradient Flow with Unrolled Highway Expectation Maximization
AAAI 2021 · Table 4
Optimizers
887.9FreeDA (ViT-L)
DINOv2 ViT-L/14 + CLIP ViT-L/14
Training-Free Open-Vocabulary Segmentation with Offline Diffusion-Augmented Prototype Generation
CVPR 2024 · Table 1
Open-set and few-shot detection
987NeCo
DINOv2R ViT-S/14
Near, far: Patch-ordering enhances vision foundation models' scene understanding
ICLR 2025 · Table: Evaluation of Full Fine
Open-set and few-shot detection
1085.4ReCLIP++
CLIP
ReCLIP++: Learn to Rectify the Bias of CLIP for Unsupervised Semantic Segmentation
IJCV 2025 · Table 1
CNN features
34 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

Pascal VOC · mIoU (val)

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

#mIoU ↑MethodPaperTopic
189.8Seg4Diff + MAGNET (COCO)
SD3
Seg4Diff: Unveiling Open-Vocabulary Semantic Segmentation in Text-to-Image Diffusion Transformers
NeurIPS 2025 · Table 1
Text-to-image diffusion
289.48Ours
DINOv2
Tunable Soft Equivariance with Guarantees
CVPR 2026 · Table 4
CNN features
385.9PixelCLIP
CLIP ViT-B/16
Towards Open-Vocabulary Semantic Segmentation Without Semantic Labels
NeurIPS 2024 · Table 1
Open-set and few-shot detection
482Ours + K-Means
Stable Diffusion 1.4
Deciphering 'What' and 'Where' Visual Pathways from Spectral Clustering of Layer-Distributed Neural Representations
CVPR 2024 · Table 1
RNNs and normalization
579.67Proposed decoder, mfb1u16c3 / 16
Xception-65
Decoders Matter for Semantic Segmentation: Data-Dependent Decoding Enables Flexible Feature Aggregation
CVPR 2019 · Table: Comparison with the van
CNN features
679.5SSR (Ours)SSR: Semantic and Spatial Rectification for CLIP-based Weakly Supervised Segmentation
AAAI 2026 · Table 1
Weakly supervised detection and segmentation
778.51ResNeXt101+Elastic
ResNeXt101
ELASTIC: Improving CNNs With Dynamic Scaling Policies
CVPR 2019 · PASCAL VOC results table
CNN design
878.4ExCEL
CLIP ViT-B
Exploring CLIP's Dense Knowledge for Weakly Supervised Semantic Segmentation
CVPR 2025 · Table 1
CLIP and zero-shot
978.3NMF with one-step gradient
ResNet-50
Is Attention Better Than Matrix Decomposition?
ICLR 2021 · Table: One-Step Gradient & BPT
Attention and transformers
1078Ours (arch0)
MobileNet-v2
Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells
CVPR 2019 · PASCAL VOC results table
Neural architecture search
16 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

Pascal VOC · mAP

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

#mAP ↑MethodPaperTopic
189.8Semantic softmax + KD
TResNet-M
ImageNet-21K Pretraining for the Masses
NeurIPS 2021 · Table, Comparing downstream re
Attention and transformers
285.75OCB (RCNN)
Mask-RCNN + SpLiCE
Object-Centric Concept-Bottlenecks
NeurIPS 2025 · Main results table
Open-set and few-shot detection
381.9Ours
Faster R-CNN / ResNet-50
Towards Unified INT8 Training for Convolutional Neural Network
CVPR 2020 · PASCAL VOC results table
Quantization
481.6Scale-aware AutoAug
Faster R-CNN ResNet-50
Scale-Aware Automatic Augmentation for Object Detection
CVPR 2021 · Table: Improvements on PASCAL
CNN features
581.02AdaBelief
Faster-RCNN
AdaBelief Optimizer: Adapting Stepsizes by the Belief in Observed Gradients
NeurIPS 2020 · Table 5
Optimizers
680.7DecentLaM
Faster R-CNN R50-FPN
DecentLaM: Decentralized Momentum SGD for Large-Batch Deep Training
ICCV 2021 · Table 5
Distributed and compressed
780.5DmSGD + one-peer exp.
Faster RCNN
Exponential Graph is Provably Efficient for Decentralized Deep Training
NeurIPS 2021 · Table 5
Graph algorithms and clustering
879Ours
Faster-RCNN, ResNet-50
Communication Efficient SGD via Gradient Sampling With Bayes Prior
CVPR 2021 · Table 5
Distributed and compressed
978.2ErfAct
SSD300
ErfAct and Pserf: Non-monotonic Smooth Trainable Activation Functions
AAAI 2022 · Table: Pascal VOC object detec
RNNs and normalization
1077.9EfficientDet-D0 + DPL (5 iter.)
EfficientDet-D0
Deep Perturbation Learning: Enhancing the Network Performance via Image Perturbations
ICML 2023 · Table 3
Attacks on text and graphs
2 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

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