---
title: 'LSFDNet: Fusion for Maritime Ship Detection'
url: https://www.emergentmind.com/topics/lsfdnet
type: topic
---

# LSFDNet: Fusion for Maritime Ship Detection

Searching arXiv for LSFDNet and closely related naming variants to ground the article in the relevant papers.
LSFDNet denotes the **Long-Short Wave Fusion Detection Network**, a **single-stage fusion and detection network** for maritime ship detection from registered short-wave infrared (SWIR) and long-wave infrared (LWIR) image pairs. It was introduced for complex nearshore conditions involving low illumination or nighttime, haze, fog, smoke, sea-surface clutter, complex backgrounds, and weak, small, or partially occluded ship targets. The method jointly optimizes image fusion and object detection in one end-to-end model, with explicit feature interaction between the two subtasks: a preliminary fused feature supplements multimodal detection, while object-sensitive detection features are fed back into fusion so that the fused image retains ship semantics more strongly [2507.20574].

## 1. Terminology and name disambiguation

The name **LSFDNet** is unambiguous in the 2025 maritime-imaging paper, where it expands to **Long-Short Wave Fusion Detection Network**. The surrounding literature, however, contains several similar acronyms that refer to different methods and tasks. This has produced recurring confusion, especially because an earlier super-resolution paper introduced **LSFD** but not an official architecture called **LSFDNet** [2507.20574].

| Term | Expansion in the cited paper | Meaning |
|---|---|---|
| **LSFDNet** | Long-Short Wave Fusion Detection Network | Official name of the SWIR–LWIR ship fusion/detection model |
| **LSFD** | Local-Selective Feature Distillation | Training/distillation method for SISR, not an official network name |
| **LSDNet** | Trainable Modification of LSD Algorithm for Real-Time Line Segment Detection | Unrelated line segment detector |
| **LS-DFN** | Dynamic Filtering with Large Sampling Field | Unrelated dynamic filtering module |

In particular, **“Local-Selective Feature Distillation for Single Image Super-Resolution”** does **not** introduce an official super-resolution architecture named **LSFDNet**. It introduces **LSFD** as a teacher-student distillation method applied to standard backbones such as RCAN, EDSR, and SAN, with only the student network used at inference [2111.10988]. Similarly, **“LSDNet: Trainable Modification of LSD Algorithm for Real-Time Line Segment Detection”** is a line detection method rather than a fusion-detection architecture for maritime imaging [2209.04642], and **“Dynamic Filtering with Large Sampling Field for ConvNets”** uses the name **LS-DFN**, not LSFDNet [1803.07624].

## 2. Operational setting and sensing rationale

LSFDNet is designed for **ship detection** in coastal scenes where single-modality imaging is unreliable. The motivating conditions include low illumination or nighttime, haze, fog, smoke, sea-surface clutter such as waves, glare, and highlights, complex nearshore backgrounds, and weak or partially occluded ships. The central premise is that **SWIR** and **LWIR** are complementary sensing bands rather than interchangeable ones [2507.20574].

The paper characterizes **SWIR** as reflective imaging, roughly **0.9–2.5 \(\mu m\)** in general, with the dataset camera operating at **0.9–1.7 \(\mu m\)**. SWIR provides better fog and smoke penetration than visible imagery, strong ship-to-sea contrast because seawater absorbs SWIR and appears dark, and richer texture and detail similar to grayscale visible images. Its weakness is dependence on illumination, especially solar irradiance; target brightness decreases under overcast, dusk, or night conditions. By contrast, **LWIR**, at **8–14 \(\mu m\)**, is thermal imaging. It is robust to illumination changes and supports stable target visibility in low light and at night, but typically lacks fine texture and edge detail and can suffer from thermal diffusion or noisy surface appearance.

This complementarity drives the model design. SWIR contributes texture, contours, and high sea-background contrast; LWIR contributes thermal saliency and illumination robustness. LSFDNet therefore treats fusion and detection as coupled tasks rather than as a conventional cascade in which a fused image is produced first and a detector is trained later. The paper explicitly frames this as an alternative to two-stage pipelines that can prioritize visual quality over downstream utility and limit information exchange between subtasks [2507.20574].

## 3. Architecture and cross-task information flow

LSFDNet takes a registered SWIR image \(I_{SW}\) and a registered LWIR image \(I_{LW}\) as input and consists of three principal components: **Multi-Task Feature Extraction (MTFE)**, the **Multi-Level Cross-Fusion (MLCF) branch**, and an **object detection branch**. The architecture is single-stage because it performs feature extraction once in a shared coupled system and trains fusion and detection jointly [2507.20574].

Within **MTFE**, the **Base Feature Extractor** applies **three \(3\times3\) convolution layers with stride 1** to each modality and outputs **8-channel shallow features** \(F_{SW}\) and \(F_{LW}\). These shallow features are shared by both tasks. The **Fusion Feature Extractor** then expands channels from **8 to 32** and reduces them back to **8**, producing \(\tilde F_{SW}\) and \(\tilde F_{LW}\). This channel expansion-and-reduction is intended to capture finer-grained features while preserving spatial resolution, which the paper treats as important because fusion is performed at pixel level.

The **detection stream** uses a **YOLOv12 backbone** and is described as having four parts: **Shallow Feature Extraction**, **Fusion Feature Augmentation**, **Deep Feature Extraction**, and **Multimodal Feature Aggregation**. The paper also states that the detector uses **A2C2f blocks** from YOLOv12, with **Area Attention** to select important features from multimodal maps. The detection branch receives \(F_{SW}\), \(F_{LW}\), and a preliminary fused feature \(F_f\), and outputs multiscale detection head features \(\{F_{head\_1}, F_{head\_2}, F_{head\_3}\}\) together with an aggregated detection attention feature \(F_{det\_attn}\).

The **MLCF branch** is the fusion core. It is composed of **three Multi-Feature Attention (MFA) blocks** and performs aggregation **across modalities, across scales, and across tasks**. At high resolution, **MFA-1** fuses \(\tilde F_{SW}\) and \(\tilde F_{LW}\) to produce \(F_{f\_H}\). Downsampled features are processed by **MFA-2**, then upsampled to yield \(F_{f\_L}\). **MFA-3** combines \(F_{f\_H}\) and \(F_{f\_L}\) into the preliminary fused feature \(F_f\). This multiscale design is then extended to multitask fusion by residually injecting \(F_{det\_attn}\) into the fusion representation, after which a fusion decoder reconstructs the final fused image \(I_{f \leftarrow det}\) [2507.20574].

The internal operation of **MFA** is patch-based. Features are divided into \(p\times p\) patches and vectorized; each modality first undergoes local self-attention, and then bidirectional cross-attention is used so that SWIR queries LWIR thermal information while LWIR queries SWIR structural information. After attention, the patch features are folded back, concatenated, and passed to a local decoder block consisting of **four convolution layers**, expanding channels from **8 to 16** and reducing them back to **8**. The paper’s interpretation is that this design preserves complementary fine and thermal cues, captures information at multiple scales, and introduces object-sensitive semantics from detection into the fusion process.

The information flow is therefore explicitly bidirectional. **Fusion \(\rightarrow\) Detection** occurs when \(F_f\) is sent back into the detection branch through **Fusion Feature Augmentation**. **Detection \(\rightarrow\) Fusion** occurs when \(F_{det\_attn}\) is residually injected into the fusion branch so that the fused representation focuses more on ships than on sea clutter. The paper presents this as the defining difference between LSFDNet and loose fusion-then-detection coupling [2507.20574].

## 4. Objective functions and optimization

LSFDNet is trained with a joint objective
\[
L = (1-\lambda)L_f + \lambda L_{det},
\]
where \(L_f\) is the fusion loss and \(L_{det}\) is the **YOLO detection loss function**. The paper reports the values \(\sigma = 0.2\), \(\alpha = 0.5\), and \(\beta = 0.5\), but **does not report the numerical value of \(\lambda\)** [2507.20574].

A distinctive component is the **Object Enhancement (OE) loss**, introduced because conventional fusion losses can preserve irrelevant sea-surface clutter as readily as ship content. The fusion loss is
\[
L_f = (1-\sigma)L_f^{global} + \sigma L_f^{object}.
\]
The **global** term is
\[
L_f^{global} = (1-\alpha)L_{global}^{grad} + \alpha L_{global}^{intensity},
\]
with
\[
L_{global}^{grad} = \frac{1}{HW}\left\| \nabla I_f - \max(\nabla I_{SW}, \nabla I_{th}) \right\|_1,
\]
\[
L_{global}^{intensity} = \frac{1}{HW}\left\| I_f - \max(I_{SW}, I_{th}) \right\|_1,
\]
and
\[
I_{th} = \mathrm{mean}(I_{SW}, I'_{LW}).
\]
Before this global term is defined, the LWIR image is gamma-corrected as
\[
I'_{LW} = 255 \times \left(\frac{I_{LW}}{255}\right)^\gamma.
\]
The paper states that this gamma correction is intended to increase ship-background contrast and suppress sea-surface noise in LWIR [2507.20574].

The **object-focused** term is
\[
L_f^{object} = (1-\beta)L_{object}^{grad} + \beta L_{object}^{intensity},
\]
where
\[
L_{object}^{grad} = \frac{1}{n}\sum_{i=1}^{n}\frac{1}{H_iW_i}\left\| \nabla I_f^i - \max(\nabla I_{SW}^i,\nabla I_{LW}^i) \right\|_1,
\]
\[
L_{object}^{intensity} = \frac{1}{n}\sum_{i=1}^{n}\frac{1}{H_iW_i}\left\| I_f^i - \max(I_{SW}^i,I_{LW}^i) \right\|_1.
\]
Here \(n\) is the number of objects, \(i\) indexes object regions, and \(H_i, W_i\) are the object-region sizes. The paper emphasizes that the object term uses the **original LWIR**, not the gamma-corrected LWIR, because ship crops exclude sea clutter and the original thermal information is treated as more trustworthy there. In effect, OE loss splits fusion into **global scene fidelity** and **local ship-focused preservation**.

The reported training setup uses **Adam**, a **learning rate of \(1\times10^{-4}\)**, **linear decay**, **500 warm-up iterations**, **30,000 total iterations**, and **batch size 8**, with **NVIDIA GeForce RTX 4090 GPUs** and **end-to-end joint optimization**. The paper does **not** report separate pretraining, exact data augmentation, optimizer betas, weight decay, mixed precision, FPS, parameter count, FLOPs, or memory use [2507.20574].

## 5. Datasets, benchmarks, and empirical behavior

The principal dataset introduced with LSFDNet is **NSLSR**, the **Nearshore Ship Long-Short Wave Registration Dataset**, created for SWIR–LWIR fusion and detection. NSLSR contains **1,205 registered SWIR–LWIR image pairs** and **2,818 annotated ship objects**. The acquisition system uses a **SWIR camera** with resolution **\(640\times512\)**, an **uncooled InGaAs FPA detector**, **\(15\,\mu m\)** pixel pitch, and **\(0.9\)–\(1.7\,\mu m\)** spectral response, together with an **LWIR camera** of the same spatial resolution, an **uncooled VOx FPA detector**, **\(12\,\mu m\)** pixel pitch, and **\(8\)–\(14\,\mu m\)** spectral response. Registration is performed by manual correction for rigid transformations, additional alignment for soft deformations using **MINIMA**, and rejection of poorly registered pairs [2507.20574].

The paper also uses **ISD** for fusion evaluation. ISD contains **1,044 image pairs** at resolution **\(300\times300\)**, but only **28 unique ship instances**, with long-range scenes and a monotonous background. For this reason, detection evaluation is performed on **NSLSR only**. The paper notes an internal inconsistency in the NSLSR split description: the dataset section says **9:1 train/test**, whereas the experiments use **844 training images** and **361 testing images**. Among the NSLSR test set, **118 images** are used for fusion evaluation and **all 361** for detection evaluation [2507.20574].

Fusion is evaluated by **EN**, **SF**, **SD**, **SCD**, **VIF**, and **Q\(_{abf}\)**. On **NSLSR**, LSFDNet reports **EN = 7.181**, **SF = 21.022**, **SD = 64.723**, **SCD = 1.427**, **VIF = 0.611**, and **Q\(_{abf}\) = 0.520**, ranking first on five metrics and second on EN. On **ISD**, it reports **EN = 7.173**, **SF = 12.330**, **SD = 50.340**, **SCD = 1.687**, **VIF = 0.848**, and **Q\(_{abf}\) = 0.651**. The qualitative discussion attributes the method’s advantage to better suppression of sea-surface noise and better preservation of ship details and brightness structure, particularly in cluttered nearshore scenes and difficult lighting [2507.20574].

Detection is evaluated by **precision**, **recall**, **mAP\(_{50}\)**, and **mAP\(_{50:95}\)**. The comparison includes single-modal baselines, fused-image detection baselines, and specialized multimodal detectors.

| Method | mAP\(_{50}\) | mAP\(_{50:95}\) |
|---|---:|---:|
| SWIR | 0.942 | 0.666 |
| LWIR | 0.929 | 0.628 |
| CAFF-DINO | 0.958 | 0.706 |
| DEYOLO | 0.956 | 0.702 |
| **LSFDNet** | **0.962** | **0.770** |

LSFDNet also reports **\(P=0.934\)** and **\(R=0.887\)**. The largest numerical gain is on the stricter localization metric, where it improves over **CAFF-DINO** from **0.706** to **0.770**. The paper further notes a qualitative case in which a small distant ship, partially occluded by the mast of a larger nearby ship, is detected only by LSFDNet [2507.20574].

The ablation studies isolate the contributions of **MLCF**, **OE loss**, and the preliminary fused feature \(F_f\). Without \(F_f\), detection drops to **\(P=0.905\)**, **\(R=0.854\)**, **\(mAP_{50}=0.953\)**, and **\(mAP_{50:95}=0.725\)**, versus **\(0.962\)** and **\(0.770\)** for the full model. For fusion, the comparison between a plain baseline and the full model shows substantial gains: from **EN 6.687, SF 18.365, SD 58.268, SCD 1.023, VIF 0.566, Qabf 0.431** to **EN 7.181, SF 21.022, SD 64.723, SCD 1.427, VIF 0.611, Qabf 0.520**. The paper explicitly states that **MLCF is critical**, **OE loss helps**, **multimodal aggregation is especially important**, and **cross-task interaction helps fusion** [2507.20574].

## 6. Relation to adjacent work, recurring misconceptions, and limitations

LSFDNet is positioned against two prior paradigms. The first is the **two-stage fusion-then-detection pipeline**, in which fusion is optimized first and a detector is trained on fused outputs later. The paper argues that such designs can increase training complexity, prioritize image appearance over detection utility, and provide little semantic feedback from detection to fusion. The second is the broader **visible–infrared fusion** literature. LSFDNet is presented as distinctive because it addresses **SWIR + LWIR**, not visible + infrared, and because the sensing physics of these bands differ materially: SWIR is reflective and detail-rich but illumination-dependent, while LWIR is emissive, lighting-robust, and texture-poor [2507.20574].

A frequent misconception is to identify LSFDNet with the super-resolution method **LSFD**. That is not the terminology used in the super-resolution paper. **“Local-Selective Feature Distillation for Single Image Super-Resolution”** introduces **LSFD** as a distillation objective and mechanism inside a teacher-student framework; it explicitly does **not** introduce an official architecture called **LSFDNet**. In that work, the underlying networks are standard backbones such as **RCAN**, **EDSR**, and **SAN**, the teacher and distillation modules are discarded at test time, and only the student backbone remains [2111.10988]. Likewise, LSFDNet should not be conflated with **LSDNet**, which is a hybrid CNN-plus-classical detector for line segments [2209.04642], or with **LS-DFN**, which is a dynamic filtering module for ConvNets [1803.07624].

The limitations reported or implied in the LSFDNet paper are domain- and implementation-specific. The method assumes **registered SWIR/LWIR input**, and the authors note that the dataset itself required manual correction and additional registration; performance may therefore degrade when misalignment is significant. The model is tuned for **maritime ship detection** and sea-background suppression, so generalization beyond that domain is not established. The paper omits several implementation specifics and does not report runtime, parameter count, FLOPs, or memory use. NSLSR fills an important data gap, but its scale remains moderate relative to large object-detection datasets, and the paper contains a split inconsistency between the stated **9:1** train/test ratio and the explicit **844/361** counts used experimentally [2507.20574].

Source: https://www.emergentmind.com/topics/lsfdnet