| PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers | ✓ Link | 80.6% | 31.1(3090) | 32.2 | PIDNet-L | 2022-06-04 |
| Semantic Flow for Fast and Accurate Scene Parsing | ✓ Link | 80.4% | 25.7(1080Ti) | 39.2 | SFNet-R18 | 2020-02-24 |
| PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers | ✓ Link | 79.8% | 42.2(3090) | 23.7 | PIDNet-M | 2022-06-04 |
| PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers | ✓ Link | 78.6% | 93.2(3090) | 10.7 | PIDNet-S | 2022-06-04 |
| Rethinking Dilated Convolution for Real-time Semantic Segmentation | ✓ Link | 78.3% | 30 | 33 | RegSeg (no ImageNet pretraining) | 2021-11-18 |
| PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model | ✓ Link | 77.5% | 102.6(1080Ti) | | PP-LiteSeg-B2 | 2022-04-06 |
| Deep Dual-resolution Networks for Real-time and Accurate Semantic Segmentation of Road Scenes | ✓ Link | 77.4% | 101.6(2080Ti) | 9.8 | DDRNet-23-slim | 2021-01-15 |
| Rethinking BiSeNet For Real-time Semantic Segmentation | ✓ Link | 76.8% | 97.0(1080Ti) | | STDC2-75 | 2021-04-27 |
| SwinMTL: A Shared Architecture for Simultaneous Depth Estimation and Semantic Segmentation from Monocular Camera Images | ✓ Link | 76.41% | | | SwinMTL | 2024-03-15 |
| HarDNet: A Low Memory Traffic Network | ✓ Link | 75.9% | 53
(1080Ti) | 18.8
(1080Ti) | U-HarDNet-70 | 2019-09-03 |
| HyperSeg: Patch-wise Hypernetwork for Real-time Semantic Segmentation | ✓ Link | 75.8% | 36.9 | 27.1 | HyperSeg-M | 2020-12-21 |
| In Defense of Pre-trained ImageNet Architectures for Real-time Semantic Segmentation of Road-driving Images | ✓ Link | 75.5% | 39.9 | | SwiftNetRN-18 | 2019-03-20 |
| Rethinking BiSeNet For Real-time Semantic Segmentation | ✓ Link | 75.3% | 126.7 | | STDC1-75 | 2021-04-27 |
| BiSeNet V2: Bilateral Network with Guided Aggregation for Real-time Semantic Segmentation | ✓ Link | 75.3% | 47.3 | 21.1 | BiSeNet V2-Large | 2020-04-05 |
| Temporally Distributed Networks for Fast Video Semantic Segmentation | ✓ Link | 74.9% | 47.6 (Titan X) | 21 | TD4-BISE18 | 2020-04-03 |
| PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model | ✓ Link | 74.9% | 143.6(1080Ti) | | PP-LiteSeg-T2 | 2022-04-06 |
| ShelfNet for Fast Semantic Segmentation | ✓ Link | 74.8% | 59.2 | 16.9 | ShelfNet18 | 2018-11-27 |
| BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation | ✓ Link | 74.7% | 65.5 | 15.2 | BiSeNet(ResNet-18) | 2018-08-02 |
| BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation | ✓ Link | 74.7% | 65.5 | | BiSeNet | 2018-08-02 |
| PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model | ✓ Link | 73.9% | 195.3(1080Ti) | | PP-LiteSeg-B1 | 2022-04-06 |
| SegBlocks: Block-Based Dynamic Resolution Networks for Real-Time Segmentation | ✓ Link | 73.8% | 48.6 (1080Ti) | | SegBlocks-RN18 (t=0.4) | 2020-11-24 |
| Rethinking BiSeNet For Real-time Semantic Segmentation | ✓ Link | 73.4% | 188.6 | | STDC2-50 | 2021-04-27 |
| BiSeNet V2: Bilateral Network with Guided Aggregation for Real-time Semantic Segmentation | ✓ Link | 72.6% | 156 | 6.4 | BiSeNet V2 | 2020-04-05 |
| PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model | ✓ Link | 72.0% | 273.6(1080Ti) | | PP-LiteSeg-T1 | 2022-04-06 |
| Rethinking BiSeNet For Real-time Semantic Segmentation | ✓ Link | 71.9% | 250.4(1080Ti) | | STDC1-50 | 2021-04-27 |
| Full-Resolution Residual Networks for Semantic Segmentation in Street Scenes | ✓ Link | 71.8% | 2.1 | 469 | FRRN | 2016-11-24 |
| []() | | 71.5% | 163.9 | 6.1 | FasterSeg | |
| Incorporating Luminance, Depth and Color Information by a Fusion-based Network for Semantic Segmentation | ✓ Link | 71.3% | 18.4 (1080Ti) | | LDFNet | 2018-09-24 |
| ESNet: An Efficient Symmetric Network for Real-time Semantic Segmentation | ✓ Link | 70.7% | 63 | 16 | ESNet | 2019-06-24 |
| LEDNet: A Lightweight Encoder-Decoder Network for Real-Time Semantic Segmentation | ✓ Link | 70.6% | 71 | 14 | LEDNet | 2019-05-07 |
| ICNet for Real-Time Semantic Segmentation on High-Resolution Images | ✓ Link | 70.6% | 30.3 | 33 | ICNet | 2017-04-27 |
| BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation | ✓ Link | 68.4% | 105.8 | 9.5 | BiSeNet(Xception39) | 2018-08-02 |
| Template-Based Automatic Search of Compact Semantic Segmentation Architectures | ✓ Link | 67.8% | 10 | 97 | Template-Based-NAS-arch1 | 2019-04-04 |
| Template-Based Automatic Search of Compact Semantic Segmentation Architectures | ✓ Link | 67.7% | 19 | 52 | Template-Based-NAS-arch0 | 2019-04-04 |
| Efficient Dense Modules of Asymmetric Convolution for Real-Time Semantic Segmentation | ✓ Link | 67.3 | 108.7 (1080Ti) | 9.2 | EDANet | 2018-09-17 |
| The Lovász-Softmax loss: A tractable surrogate for the optimization of the intersection-over-union measure in neural networks | ✓ Link | 63.1% | 76.9 | 13 | ENet + Lovász-Softmax | 2017-05-24 |
| Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs | ✓ Link | 63.1% | 0.25 | 4000 | DeepLab | 2014-12-22 |
| Conditional Random Fields as Recurrent Neural Networks | ✓ Link | 62.5% | 1.4 | 700 | CRF-RNN | 2015-02-11 |
| ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation | ✓ Link | 58.3% | 76.9 | 13 | ENet | 2016-06-07 |