Semantic segmentation demands a careful balance between accuracy, efficiency, and scalability, which remains difficult to achieve for high-resolution imagery. Convolutional networks effectively model local patterns but struggle with long-range dependencies, whereas Vision Transformers capture global context at a high computational cost. While recent work largely focuses on encoder design, the bottleneck stage, central to contextual aggregation and information flow, has been relatively overlooked. We propose SiConMo, a lightweight yet effective framework, implemented in two variants: an RGB-only model (SiConMo) and a GME-enhanced variant (SiConMo†). We show that simplicity arises from a key design principle: at very low computational budgets, the bottleneck is the most efficient stage to integrate local and global context. SiConMo integrates three complementary components: a Token Pyramid Extraction Module for hierarchical multi-scale representation, a Transformer-Branched Depthwise Convolution block for bottleneck-aware context modeling, and a Feature Merging Module that preserves spatial structure while enhancing semantic consistency. Extensive experiments on ADE20K, PASCAL Context, Cityscapes, and COCO-Stuff demonstrate that SiConMo achieves a state-of-the-art accuracy-efficiency trade-off among lightweight semantic segmentation models, highlighting simplicity as a powerful design principle.
Figures & tables
Figure 1: (Top) Efficiency–accuracy comparison of lightweight semantic segmentation models on ADE20K validation set. Circle sizes denote GFLOPs. SiConMo achieves a superior trade-off, illustrating the effectiveness of a simple bottleneck-aware design principle. (Bottom) Variants of the proposed SiConMo architecture.
Figure 2: Overview of the proposed SiConMo framework. The architecture integrates the Token Pyramid Extraction Module (TPEM) for multi-scale representation, the Trans-BDC block for bottleneck-aware context modeling, and the Feature Merging Module (FMM) for adaptive spatial–semantic integration.
Models
Input
mIoU
GFLOPs
Params
Latency(ms)
SiConMo*
448×448
34.4
0.5
1.7M
14
SiConMo † *
448×448
34.7
0.5
1.7M
14
Table 1: Quantitative evaluation on ADE20K [ 32 ] and complementary analyses. (a) Comparison with state-of-the-art representative semantic segmentation models on ADE20K dataset. (b) Input-resolution variants trained at 448×448 . (c) Ablation study on ADE20K. (d) GME variants on ADE20K. GFLOPs are reported at 512×512 input resolution with single-scale inference; ‘-’ denotes unreported values. (e) object detection results on COCO dataset using RetinaNet (Sec. 4.3 ). SiConMo † : GME variant.
Figure 3: Visual results on the ADE20K validation set. The proposed SiConMo model produces high-quality segmentation maps with improved spatial consistency and fine-detail preservation, compared to TopFormer [ 30 ] .
Methods
Backbone
mIoU 59
mIoU 60
GFLOPs
DeepLabV3+ [ 3 ]
ENet-s16 [ 22 ]
43.07
39.19
23.00
DeepLabV3+ [ 3 ]
MobileNetV2-s16 [ 23 ]
42.34
38.59
22.24
LR-ASPP [ 11 ]
MobileNetV3-s16 [ 11 ]
38.02
35.05
2.04
TopFormer [ 30 ]
TopFormer-T [ 30 ]
40.39
36.41
0.53
SeaFormer [ 26 ]
SeaFormer-T [ 26 ]
41.49
37.27
0.51
SiConMo (Ours)
Ours
41.84
37.49
0.47
Table 2: Results on PASCAL Context test set [ 19 ] .
Methods
Encoder
mIoU
GFLOPs
PSPNet [ 31 ]
MobileNetV2 [ 23 ]
70.2
423.4
FCN [ 16 ]
MobileNetV2 [ 23 ]
61.5
317.1
SegFormer [ 27 ]
MiT-B0 [ 27 ]
71.9
17.7
L-ASPP [ 3 ]
MobileNetV2 [ 23 ]
72.7
12.6
LR-ASPP [ 11 ]
MobileNetV3-Large [ 11 ]
72.4
9.7
LR-ASPP [ 11 ]
MobileNetV3-Small [ 11 ]
68.4
2.9
Table 3: Results on Cityscapes validation set [ 4 ] .
Methods
Encoder
mIoU
GFLOPs
PSPNet [ 31 ]
MobileNetV2-s8 [ 23 ]
30.14
52.94
DeepLabV3+ [ 3 ]
EfficientNet-s16 [ 25 ]
31.45
27.10
DeepLabV3+ [ 3 ]
MobileNetV2-s16 [ 23 ]
29.88
25.90
LR-ASPP [ 11 ]
MobileNetV3-s16 [ 11 ]
25.16
2.37
TopFormer [ 30 ]
TopFormer-T [ 30 ]
28.34
0.64
SeaFormer [ 26 ]
SeaFormer-T [ 26 ]
29.24
0.62
Table 4: Results on COCO-Stuff test set [ 1 ] .
Figure 4: The architecture of the proposed SiConMo model for the task of image classification.
Method
Input Size
Top-1 Accuracy(%)
GFLOPs
Parameters
SiConMo (Ours)
224×224
66.2
0.13
1.79M
SiConMo † (Ours)
224×224
66.4
0.13
1.79M
Table 5: SiConMo results for ImageNet classification.
Stage
Details
Output Resolution
layer
kernel size
expand ratio
output channels
stride
Stem
Conv
3
-
16
2
MobileNetV2
3
1
16
1
256×256
MobileNetV2
3
4
16
2
MobileNetV2
3
3
16
1
128×128
MobileNetV2
5
3
32
2
Table 6: Architectural details of SiConMo model. For input resolution of 512×512 .
Figure 5: Design of Feed-Forward Network.
Figure 6: Visual results on the ADE20K validation set. Our proposed model illustrates superior consistency with the ground truth segmentation results, highlighting its robustness.
Modern vision models must capture image-level context without sacrificing local detail while remaining computationally affordable. We revisit this tradeoff and advance a simple principle: decouple the roles of global reasoning and local representation. To operationalize this principle, we introduce ConvNeur, a two-branch architecture in which a lightweight neural memory branch aggregates global context on a compact set of tokens, and a locality-preserving branch extracts fine structure. A learned gate lets global cues modulate local features without entangling their objectives. This separation yields subquadratic scaling with image size, retains inductive priors associated with local processing, and reduces overhead relative to fully global attention. On standard classification, detection, and segmentation benchmarks, ConvNeur matches or surpasses comparable alternatives at similar or lower compute and offers favorable accuracy versus latency trade-offs at similar budgets. These results support the view that efficiency follows global-local decoupling.
Zhenyu Yang, Gensheng Pei, Tao Chen +4
Nanjing University of Science and Technology · State Key Laboratory of Intelligent Manufacturing of Advanced Construction Machinery · Department of Electrical and Computer Engineering, Sungkyunkwan University +2
Recent lightweight semantic segmentation methods have made significant progress by combining compact backbones with efficient decoder heads. However, most multi-scale decoders compute attention independently at each feature scale, introducing substantial redundancy since the resulting attention distributions across scales are strongly correlated. We propose Cross-Stage Attention Propagation (CSAP), a decoder framework that computes attention at the deepest feature scale and propagates the resulting attention maps to shallower stages, bypassing query-key computation at those stages entirely. This design preserves multi-scale contextual reasoning while substantially reducing the decoder's computational cost. CSAP-Tiny achieves 42.9% mIoU on ADE20K with only 5.5 GFLOPs, 80.5% on Cityscapes with 21.5 GFLOPs, and 40.9% on COCO-Stuff 164K with 5.5 GFLOPs, surpassing SegNeXt-Tiny by +1.8% on ADE20K while requiring 16.8% fewer floating-point operations.
This paper presents Focus-on-Regions Network (FoR-Net), an efficient semantic segmentation framework that explicitly focuses on hard regions through a selector-driven Top-K mechanism. Instead of relying on heavy global modeling, FoR-Net selectively enhances structurally informative regions. Multi-scale reasoning branches are introduced to aggregate spatial context efficiently. Experiments on the Cityscapes dataset demonstrate that FoR-Net achieves competitive performance while maintaining a lightweight architecture.
Sheng-Wei Chan, Hsin-Jui Pan, Jen-Shiun Chiang
Department of Electrical and Computer Engineering, Tamkang University, No.151, Yingzhuan Rd., Tamsui Dist., New Taipei City, 251301, Taiwan