paper-with-me

Semantic Segmentation 벤치마크

Semantic Segmentation on DensePASS

72개 결과 · ⬇ CSV · JSON

mIoU

16.65 26.79 36.94 47.08 57.23 2016-12 2026-09 PSPNet (ResNet-50) — 29.5 (2016-12-04) PSPNet (ResNet-50) — 29.5 (2016-12-04) ERFNet — 16.65 (2017-10-09) ERFNet — 16.65 (2017-10-09) DeepLabV3+ (ResNet-101) — 32.5 (2018-02-07) DeepLabV3+ (ResNet-101) — 32.5 (2018-02-07) DANet (ResNet-101) — 28.5 (2018-09-09) DANet (ResNet-101) — 28.5 (2018-09-09) CLAN — 31.46 (2018-09-25) CLAN — 31.46 (2018-09-25) USSS (Mapillary) — 30.87 (2018-11-26) USSS (IDD) — 26.98 (2018-11-26) USSS (Mapillary) — 30.87 (2018-11-26) USSS (IDD) — 26.98 (2018-11-26) Semantic-FPN (ResNet-101) — 28.8 (2019-01-08) Semantic-FPN (ResNet-101) — 28.8 (2019-01-08) Fast-SCNN — 24.6 (2019-02-12) Fast-SCNN — 24.6 (2019-02-12) SwiftNet (Cityscapes) — 25.67 (2019-03-20) SwiftNet (Cityscapes) — 25.67 (2019-03-20) Seamless (Mapillary) — 34.14 (2019-05-03) Seamless (Mapillary) — 34.14 (2019-05-03) CRST — 31.67 (2019-08-26) CRST — 31.67 (2019-08-26) PASS — 23.66 (2019-09-17) PASS — 23.66 (2019-09-17) SIM — 44.58 (2020-03-18) SIM — 44.58 (2020-03-18) DNL (ResNet-101) — 32.1 (2020-06-11) DNL (ResNet-101) — 32.1 (2020-06-11) FANet (Resnet-34) — 26.9 (2020-07-07) FANet (Resnet-34) — 26.9 (2020-07-07) SwiftNet (Merge3) — 32.04 (2020-08-20) SwiftNet (Merge3) — 32.04 (2020-08-20) SETR (PUP, Transformer-L) — 35.7 (2020-12-31) SETR (MLA, Transformer-L) — 35.6 (2020-12-31) SETR (PUP, Transformer-L) — 35.7 (2020-12-31) SETR (MLA, Transformer-L) — 35.6 (2020-12-31) PVT (Tiny, FPN) — 31.2 (2021-02-24) PVT (Tiny, FPN) — 31.2 (2021-02-24) ECANet — 43.02 (2021-03-09) ECANet — 43.02 (2021-03-09) PCS — 53.83 (2021-03-31) PCS — 53.83 (2021-03-31) SegFormer (MiT-B2) — 42.4 (2021-05-31) SegFormer (MiT-B1) — 38.5 (2021-05-31) SegFormer (MiT-B2) — 42.4 (2021-05-31) SegFormer (MiT-B1) — 38.5 (2021-05-31) ASMLP (MiT-B1) — 42.05 (2021-07-18) ASMLP (MiT-B1) — 42.05 (2021-07-18) CycleMLP (MiT-B1) — 40.16 (2021-07-21) CycleMLP (MiT-B1) — 40.16 (2021-07-21) DPT (MiT-B1) — 36.5 (2021-07-30) DPT (MiT-B1) — 36.5 (2021-07-30) P2PDA (Cityscapes+WildDash) — 48.52 (2021-10-21) P2PDA (Cityscapes) — 41.99 (2021-10-21) P2PDA (Cityscapes+WildDash) — 48.52 (2021-10-21) P2PDA (Cityscapes) — 41.99 (2021-10-21) PoolFormer (MiT-B1) — 43.18 (2021-11-22) PoolFormer (MiT-B1) — 43.18 (2021-11-22) DAFormer — 54.67 (2021-11-29) DAFormer — 54.67 (2021-11-29) Trans4PASS (multi-scale) — 56.38 (2022-03-02) Trans4PASS (single-scale) — 55.25 (2022-03-02) Trans4PASS (multi-scale) — 56.38 (2022-03-02) Trans4PASS (single-scale) — 55.25 (2022-03-02) FAN (MiT-B1) — 42.54 (2022-04-26) FAN (MiT-B1) — 42.54 (2022-04-26) Trans4PASS+ (multi-scale) — 57.23 (2022-07-25) Trans4PASS+ (single-scale) — 56.45 (2022-07-25) Trans4PASS+ (multi-scale) — 57.23 (2022-07-25) Trans4PASS+ (single-scale) — 56.45 (2022-07-25) PSPNet (ResNet-50) — 29.5 (2016-12-04) DeepLabV3+ (ResNet-101) — 32.5 (2018-02-07) Seamless (Mapillary) — 34.14 (2019-05-03) SIM — 44.58 (2020-03-18) PCS — 53.83 (2021-03-31) DAFormer — 54.67 (2021-11-29) Trans4PASS (multi-scale) — 56.38 (2022-03-02) Trans4PASS+ (multi-scale) — 57.23 (2022-07-25)
RankModel mIoU Extra Training Data PaperCodeYear
1 Trans4PASS+ (multi-scale) 57.23% Behind Every Domain There is a Shift: Adapting Distortion-aware Vision Transformers for Panoramic Semantic Segmentation jamycheung/trans4pass 2022
2 Trans4PASS+ (single-scale) 56.45% Behind Every Domain There is a Shift: Adapting Distortion-aware Vision Transformers for Panoramic Semantic Segmentation jamycheung/trans4pass 2022
3 Trans4PASS (multi-scale) 56.38% Bending Reality: Distortion-aware Transformers for Adapting to Panoramic Semantic Segmentation jamycheung/trans4pass 2022
4 Trans4PASS (single-scale) 55.25% Bending Reality: Distortion-aware Transformers for Adapting to Panoramic Semantic Segmentation jamycheung/trans4pass 2022
5 DAFormer 54.67% DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation lhoyer/DAFormer · dbash/visda2022-org · kw01sg/crda 2021
6 PCS 53.83% Prototypical Cross-domain Self-supervised Learning for Few-shot Unsupervised Domain Adaptation zhengzangw/PCS-FUDA 2021
7 P2PDA (Cityscapes+WildDash) 48.52% ✓ Transfer beyond the Field of View: Dense Panoramic Semantic Segmentation via Unsupervised Domain Adaptation chma1024/densepass 2021
8 SIM 44.58% Differential Treatment for Stuff and Things: A Simple Unsupervised Domain Adaptation Method for Semantic Segmentation SHI-Labs/Unsupervised-Domain-Adaptation-with-Differential-Treatment 2020
9 PoolFormer (MiT-B1) 43.18% MetaFormer Is Actually What You Need for Vision huggingface/transformers · rwightman/pytorch-image-models · facebookresearch/xformers · +15 2021
10 ECANet 43.02% ✓ Capturing Omni-Range Context for Omnidirectional Segmentation elnino9ykl/WildPASS 2021
11 FAN (MiT-B1) 42.54% Understanding The Robustness in Vision Transformers nvlabs/fan · NVlabs/STL 2022
12 SegFormer (MiT-B2) 42.4% SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers huggingface/transformers · VikParuchuri/surya · PaddlePaddle/PaddleSeg · +25 2021
13 ASMLP (MiT-B1) 42.05% AS-MLP: An Axial Shifted MLP Architecture for Vision liuruiyang98/Jittor-MLP · svip-lab/AS-MLP 2021
14 P2PDA (Cityscapes) 41.99% Transfer beyond the Field of View: Dense Panoramic Semantic Segmentation via Unsupervised Domain Adaptation chma1024/densepass 2021
15 CycleMLP (MiT-B1) 40.16% CycleMLP: A MLP-like Architecture for Dense Prediction BR-IDL/PaddleViT · ShoufaChen/CycleMLP · liuruiyang98/Jittor-MLP · +5 2021
16 SegFormer (MiT-B1) 38.5% SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers huggingface/transformers · VikParuchuri/surya · PaddlePaddle/PaddleSeg · +25 2021
17 DPT (MiT-B1) 36.50% DPT: Deformable Patch-based Transformer for Visual Recognition CASIA-IVA-Lab/DPT 2021
18 SETR (PUP, Transformer-L) 35.7% Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers PaddlePaddle/PaddleSeg · BR-IDL/PaddleViT · fudan-zvg/SETR · +2 2020
19 SETR (MLA, Transformer-L) 35.6% Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers PaddlePaddle/PaddleSeg · BR-IDL/PaddleViT · fudan-zvg/SETR · +2 2020
20 Seamless (Mapillary) 34.14% ✓ Seamless Scene Segmentation mapillary/seamseg · gjp1203/LIV360SV · gladcolor/seamseg · +2 2019
1–20 / 72 다음 → 페이지당 10 20 50 100