SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.01278 · 2026

DSFC-Net: A Dual-Encoder Spatial and Frequency Co-Awareness Network for Rural Road Extraction

Lin Chen, Ying Wang, Kun Ding, Shiming Xiang, Yue Sun, Bing Xu, Zhengbo Zhang, Wanke Xia, Yihe Tian

arXiv · PDF · Open in the Atlas

Code that ran

9 of 11 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (9 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 9 have no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2602.01278.json · badge: /api/ran/2602.01278.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
ZhengboZhang/DSFC-Net community
Papers with Code archive
9 ran · 2 no recorded run
FunctionStatusWhere it lives
CFFM Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("1fbc21aeb39b27bd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, CFFM: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class CFFM(nn.Module):

    def __init__(self, channels, reduction_ratio=4):
        super(CFFM, self).__init__()
        self.channels = channels
        self.gap = nn.AdaptiveAvgPool2d(1)
        self.fc = nn.Sequential(nn.Linear(2 * channels, 2 * channels // reduction_ratio, bias=False), nn.ReLU(inplace=True), nn.Linear(2 * channels // reduction_ratio, 2 * channels, bias=False), nn.Sigmoid())

    def forward(self, feat_a, feat_b):
        assert feat_a.size(1) == feat_b.size(1) == self.channels, 'Channel dimensions must match!'
        batch_size = feat_a.size(0)
        feat_a_squeezed = self.gap(feat_a).view(batch_size, self.channels)
        feat_b_squeezed = self.gap(feat_b).view(batch_size, self.channels)
        squeezed_combined = torch.cat([feat_a_squeezed, feat_b_squeezed], dim=1)
        excitation = self.fc(squeezed_combined)
        weight_a, weight_b = torch.split(excitation, [self.channels, self.channels], dim=1)
        weight_a = weight_a.view(batch_size, self.channels, 1, 1)
        weight_b = weight_b.view(batch_size, self.channels, 1, 1)
        feat_a_weighted = feat_a * weight_a
        feat_b_weighted = feat_b * weight_b
        fused_feat = feat_a_weighted + feat_b_weighted
        return fused_feat


# Create instance and test inputs
channels = 4
reduction_ratio = 2

cffm_module = CFFM(channels=channels, reduction_ratio=reduction_ratio)

# Create feature tensors: batch_size=2, channels=4, height=3, width=5
feat_a = torch.randn(2, 4, 3, 5)
feat_b = torch.randn(2, 4, 3, 5)

ARGS = {
    'channels': channels,
    'reduction_ratio': reduction_ratio
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type CFFM — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a CFFM output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
CFIA Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("77e71806d5e4a4c3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, CFIA: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal valid input tensor
# Shape: (batch=2, channels=4, height=16, width=16)
# Height and width must be divisible by split_size=8
x = torch.randn(2, 4, 16, 16)

# Define ARGS dict with all parameters for CFIA.__init__
ARGS = {
    'dim': 4,
    'num_heads': 8,
    'split_size': 8
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type CFIA — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a CFIA output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Conv_Block Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("34a89705a067acdd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, Conv_Block: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from timm.layers import DropPath

torch.manual_seed(0)

# Create a Conv_Block instance to inspect what parameters it takes
# Conv_Block.__init__ takes: dim, drop_path=0.0

ARGS = {
    "dim": 4,
    "drop_path": 0.1,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Conv_Block — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Conv_Block output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
DW_Block Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("80a6e2f4e280a090")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, DW_Block: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "in_features": 4,
    "kernel_size": 3,
    "padding": 1,
    "hidden_features": 8,
    "out_features": 4,
    "act_layer": nn.GELU,
    "drop": 0.1,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DW_Block — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a DW_Block output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
GRN Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("1ab5d1761e67dbec")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, GRN: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# GRN only takes dim as a parameter
# dim should be an integer representing the channel/feature dimension
dim = 4

ARGS = {
    'dim': dim,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GRN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GRN output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
MFFN Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("4d5de7916ccfaae1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, MFFN: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "dim": 8,
    "mlp_ratio": 4,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MFFN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MFFN output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
SCA Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("6888463293eb6e4c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, SCA: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class SCA(nn.Module):

    def __init__(self, dim):
        super(SCA, self).__init__()
        self.dim = dim
        self.dim_sp = dim
        self.conv1 = nn.Conv2d(self.dim_sp, self.dim_sp, 3, padding=1, dilation=1, groups=self.dim_sp)
        self.conv2 = nn.Conv2d(self.dim_sp, self.dim_sp, 3, padding=2, dilation=2, groups=self.dim_sp)
        self.conv_fusion = nn.Sequential(nn.Conv2d(2 * self.dim, 2 * self.dim, kernel_size=3, stride=1, padding=1), nn.BatchNorm2d(2 * self.dim), nn.ReLU(inplace=True), nn.Conv2d(2 * self.dim, self.dim, kernel_size=3, stride=1, padding=1), nn.BatchNorm2d(self.dim), nn.ReLU(inplace=True))

    def forward(self, x):
        cd1 = self.conv1(x)
        cd2 = self.conv2(x)
        out = torch.cat([cd1, cd2], dim=1)
        out = self.conv_fusion(out)
        return out

# Create a SCA instance with dim=4
sca_model = SCA(dim=4)

ARGS = {
    "dim": 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type SCA — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a SCA output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
UpSample Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("c6dfc691dd44fc7b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, UpSample: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import torch.nn.functional as F

torch.manual_seed(0)

class LayerNorm(nn.Module):
    """ LayerNorm that supports two data formats: channels_last (default) or channels_first.
    The ordering of the dimensions in the inputs. channels_last corresponds to inputs with
    shape (batch_size, height, width, channels) while channels_first corresponds to inputs
    with shape (batch_size, channels, height, width).
    """

    def __init__(self, normalized_shape, eps=1e-06, data_format='channels_last'):
        super().__init__()
        self.weight = nn.Parameter(torch.ones(normalized_shape))
        self.bias = nn.Parameter(torch.zeros(normalized_shape))
        self.eps = eps
        self.data_format = data_format
        if self.data_format not in ['channels_last', 'channels_first']:
            raise NotImplementedError
        self.normalized_shape = (normalized_shape,)

    def forward(self, x):
        if self.data_format == 'channels_last':
            return F.layer_norm(x, self.normalized_shape, self.weight, self.bias, self.eps)
        elif self.data_format == 'channels_first':
            u = x.mean(1, keepdim=True)
            s = (x - u).pow(2).mean(1, keepdim=True)
            x = (x - u) / torch.sqrt(s + self.eps)
            x = self.weight[:, None, None] * x + self.bias[:, None, None]
            return x

class UpSample(nn.Module):

    def __init__(self, input_ch, output_ch):
        super(UpSample, self).__init__()
        self.proj = nn.Sequential(LayerNorm(input_ch, eps=1e-06, data_format='channels_first'), nn.ConvTranspose2d(input_ch, output_ch, kernel_size=2, stride=2))

    def forward(self, x):
        x = self.proj(x)
        return x

ARGS = {
    'input_ch': 4,
    'output_ch': 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type UpSample — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a UpSample output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
up_block Ran
ran on a synthesized input · constructed an object; no result was computed
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("bae0d6cd781e2a35")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, up_block: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class up_block(nn.Module):

    def __init__(self, in_ch, out_ch):
        super(up_block, self).__init__()
        self.conv = nn.Sequential(nn.Conv2d(in_ch, in_ch, kernel_size=3, stride=1, padding=1, bias=True), nn.BatchNorm2d(in_ch), nn.ReLU(inplace=True), nn.Conv2d(in_ch, out_ch, kernel_size=3, stride=1, padding=1, bias=True), nn.BatchNorm2d(out_ch), nn.ReLU(inplace=True))

    def forward(self, x):
        x = self.conv(x)
        return x

ARGS = {
    "in_ch": 4,
    "out_ch": 8,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type up_block — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a up_block output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
DSFC Not yet run
unverified (harvested, no recorded run)
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("0c7e562344eaebcd")
How it ran

no invocation recorded.

SFT Not yet run
unverified (harvested, no recorded run)
ZhengboZhang/DSFC-Net/model.py
code served (permissive licence) · get_code("8c109203353301b4")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 10 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
GRN
brick of this paper's assembly: model
1 other paper
1 dated earlier · 1 only as an assembly brick
2301.00808 (2023-01)
ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Accurate extraction of rural roads from highresolution remote sensing imagery is essential for infrastructure planning and sustainable development. However, this task presents unique challenges in rural settings due to several factors. These include high intra-class variability and low inter-class separability from diverse surface materials, frequent vegetation occlusions that disrupt spatial continuity, and narrow road widths that exacerbate detection difficulties. Existing methods, primarily optimized for structured urban environments, often underperform in these scenarios as they overlook such distinctive characteristics. To address these challenges, we propose DSFC-Net, a dual-encoder framework that synergistically fuses spatial and frequency-domain information. Specifically, a CNN branch is employed to capture fine-grained local road boundaries and short-range continuity, while a novel Spatial-Frequency Hybrid Transformer (SFT) is introduced to robustly model global topological dependencies against vegetation occlusions. Distinct from standard attention mechanisms that suffer from frequency bias, the SFT incorporates a Cross-Frequency Interaction Attention (CFIA) module that explicitly decouples high-and lowfrequency information via a Laplacian Pyramid strategy. This design enables the dynamic interaction between spatial details and frequency-aware global contexts, effectively preserving the connectivity of narrow roads. Furthermore, a Channel Feature Fusion Module (CFFM) is proposed to bridge the two branches by adaptively recalibrating channel-wise feature responses, seamlessly integrating local textures with global semantics for accurate segmentation. Comprehensive experiments on the WHU-RuR+, DeepGlobe, and Massachusetts datasets validate the superiority of DSFC-Net over state-of-the-art approaches. Notably, our method attains an F1-score of 69.93% and an IoU of 53.77% on the challenging WHU-RuR+ dataset, highlighting its robustness in rural scenes with narrow, sparse, and occluded roads. The code is available at https://github.com/ZhengboZhang/DSFC-Net.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2602.01278")
get_code_for_paper("2602.01278")
have("2602.01278")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2602.01278.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2602.01278.svg)](https://syntology.ai/paper/2602.01278)

Connect an agent — have() is free.