SCI-Mamba: Unsupervised Learning based Low-Light Image Enhancement for Non-Cooperative Spacecraft

Light the Moon: Asymmetric Physics-Guided Dual-Domain Network for Low-Light Enhancement of Lunar Shadowed Regions

Low-light image enhancement of remote sensing imagery within lunar permanently shadowed regions (PSRs) is a critical prerequisite for ensuring safe landing and precise navigation for future crewed missions and geological resource prospecting. Given the absence of direct sunlight, images acquired in PSRs typically exhibit insufficient brightness, severe noise, and degradation of structural details, which are detrimental to lunar polar oriented exploration tasks. Furthermore, constrained by the lack of paired datasets, existing methods rely on unsupervised, single-degradation paradigms; their inability to accurately model complex lunar degradations within a unified framework leads to suboptimal results. To address these challenges, we propose an asymmetric physics-guided dual-domain network (APDNet), which decomposes the PSR enhancement task into illumination decoupling, illumination restoration, and structure restoration. The asymmetric architecture allows us to perform task-oriented enhancements for different modules according to their respective subtasks. Specifically, a frequency-domain processing mechanism is introduced to achieve spectral supervision for illumination restoration. Global and local latent information is processed in parallel and fused to restore degraded textures. In addition, the first paired low-light image dataset specifically dedicated to lunar PSR is constructed in this work. Through the modeling and learning of the degradation parameters of the real-world lunar environment, a PSR synthetic dataset is also constructed to facilitate supervised model training. Extensive experiments demonstrate that our proposed model achieves state-of-the-art (SOTA) enhancement performance on the PSR low-light dataset. Furthermore, evaluations on generic low-light datasets demonstrate competitive performance and strong generalization capabilities compared to existing SOTA methods, thoroughly validating the effectiveness of our approach.

Self-Supervised Vision Transformers for Spacecraft Pose Estimation Across Domain Shift

Accurate monocular pose estimation of noncooperative spacecraft is critical for autonomous proximity operations such as on-orbit servicing and active debris removal. A major practical challenge is the domain gap between synthetic training imagery and real Hardware-In-the-Loop (HIL) sensor data. We present a pose estimation pipeline built on a self-supervised DINOv3 Vision Transformer backbone with a deconvolutional heatmap head that localizes spacecraft keypoints. To bridge the synthetic-to-real gap without target-domain annotations, the pipeline combines multi-scale structural similarity (MS-SSIM) supervision, feature-level domain generalization, input-level style randomization, and iterative self-training with pseudo-labels from unlabeled HIL images. On the SPEED+ benchmark, the pipeline achieves state-of-the-art accuracy on both HIL domains using a single model with no adversarial training, no external augmentation networks, and no target-domain annotations.

Zero-Shot Illumination and Noise Estimation for Edge-Deployable Low-Light Image Enhancement in IoT Systems

In Internet of Things (IoT) systems, such as autonomous vehicles and robots, visual perception under low-light conditions is often severely degraded due to insufficient illumination and intensive sensor noise. While numerous methods have been developed for low-light enhancement, their reliance on paired annotation and computationally heavy architectures limits their applicability for downstream perception tasks, especially on resource-constrained platforms. To address these issues, this article proposes REZS-Edge, a novel zero-shot enhancement framework that eliminates the need for paired supervision and is specifically designed for efficient edge execution. First, by integrating Retinex theory with the noise-to-noise paradigm, the method jointly estimates illumination conditions and suppresses noise through a unified pipeline. Then, a zero-shot training strategy with global and local illumination constraints facilitates fully unsupervised optimization and enhances generalization. Furthermore, the framework incorporates structural re-parameterization to enable model acceleration at the inference phase while maintaining performance. Extensive experiments demonstrate that REZS-Edge achieves state-of-the-art results on low-light image enhancement (LLIE) datasets, including LOLv1 and LOLv2-Real, with a notably fast inference speed of 6.25 ms on a commercial SoC platform Snapdragon 8 Gen 3. More importantly, the method exhibits strong generalization on critical downstream tasks such as low-light object detection, instance segmentation, and pose estimation. These advantages make REZS-Edge particularly valuable for autonomous driving and robotic systems that require reliable, efficient, and all-weather visual perception, offering a practical and deployable solution for real-time enhancement under challenging lighting conditions.

Real-time detection of outdoor non-obvious anthropogenic trace via texture contrast learning

From Sim to 6DOF: Deep Learning for Real-Time Satellite Pose Estimation from Resolved Ground-Based Imagery

This work presents the first complete system for automated six degrees of freedom (6DOF) satellite pose estimation from spatially resolved, ground-based, adaptive optics (AO)-corrected imagery, addressing a key challenge in Space Domain Awareness (SDA). The approach mitigates the need for human labeling by directly regressing satellite orientation and position from blurry, noisy, and deeply shadowed imagery. A multi-stage deep neural network pipeline localizes the satellite, predicts pose, and optionally applies temporal filtering. Networks are trained exclusively on fully synthetic imagery generated from a CAD model, yet generalize effectively to real data, bridging the Sim2Real domain gap. On 137 real, human-labeled test images of Seasat, the model achieved a mean rotation error of 5° and a mean image-plane translation error of 21 cm. Slant range error was quantitatively evaluated on synthetic data due to unknown real-sensor parameters. Qualitative evaluation of additional real Seasat imagery rated 177 of 199 predicted poses as “ground truth equivalent” or “high-confidence match,” with zero catastrophic failures. The system was extended to seven degrees of freedom (7DOF) for satellites with articulating components and demonstrated on real Hubble Space Telescope (HST) imagery, achieving 5.5° rotation error, 51 cm image-plane translation error, and 8° symmetry-adjusted solar array error on a 249-frame pass with causal temporal filtering. Across 586 real test images from Seasat and HST (captured over multiple decades under diverse conditions) the system consistently performed well. Full 6DOF performance was quantified on a high-fidelity wave optics (HFWO) synthetic test set of Seasat, where the model achieved 8.4° mean rotation error, 34 cm image-plane translation error, and 1.4% line-of-sight range error at r0=6 cm and 1031 km range. In a limited 200-image benchmark, the model demonstrated 48% lower mean rotation error than a single human labeler while operating ∼800× faster. It required <40 h and a single A100 GPU to generate data and train. The approach was also demonstrated for ARGOS, a smaller satellite with highly symmetric geometry. An exploratory General Image-Quality Equation-based image quality metric (AO-IQ) was introduced as an empirical correlate for pose accuracy. General-purpose models like GPT-4o and Depth Anything V2 failed across most SDA tasks, but rapid gains in vision-language models warrant continued monitoring. These results establish a new operational baseline for practical, real-time satellite pose estimation from AO SDA imagery.

Vernata: Self-Supervised Learning of LiDAR Point Representations

LiDAR serves as a primary sensing modality for robots operating in outdoor environments. However, the performance of deep learning models in this domain is severely limited by the scarcity of labeled data, a direct result of the high cost of 3D annotation. Self-supervised learning addresses this scarcity by learning general-purpose features from unlabeled data. In this work, we present a multi-modal, multi-teacher distillation framework for self-supervised learning on outdoor LiDAR point clouds. Building upon the Sonata architecture, we introduce Vernata, consisting of three extensions: sparse view augmentation to improve robustness against varying point densities, a memory bank mechanism to stabilize resource-constrained training, and cross-modal distillation utilizing dense, high-resolution 2D image features to enable fine-grained semantic guidance. We evaluate our method on the GrandTour, TartanGround, and Waymo datasets, as well as data collected from our own robotic platforms. Our experiments demonstrate a significant performance improvement over Sonata baselines, yielding mIoU scores of 54.7 on TartanGround (+5.9 points, +12.1%) and 57.1 on Waymo (+7.3 points, +14.7%). Finally, we show that the self-supervised approach maintains strong performance even in reduced-modality settings (lacking color or normals), achieving competitive mIoU scores of 49.4 and 50.2 on the respective datasets.