Search Results (217)

Lane detection is a core perception task for Advanced Driver Assistance Systems (ADAS) and autonomous driving. Current methods struggle to balance accuracy, model complexity and inference efficiency: high-precision models rely on heavy modules with excessive computation, while lightweight ones suffer from weak feature extraction and low precision. To alleviate this inherent trade-off, we propose YOLOv8n-LaneDG based on YOLOv8n-seg. We design a dual-path gated fusion block to strengthen lane features and an efficient upsampling convolution block to reduce computational overhead, and we further design a weighted continuity loss to preserve lane structural integrity. Evaluated on TuSimple, our method lifts mAP@0.5 from 74.3% to 95.2%. It outperforms mainstream lightweight models and matches heavy YOLOv8s-seg with far fewer parameters, delivering a high-precision, deployable lane detection solution for vehicle-end platforms. Full article

Airside vehicles operate close to aircraft, service equipment and safety-critical surface markings, yet driver-assistance functions developed for road traffic do not directly transfer to airport aprons, taxiways and service roads. This article presents a vision-based driver-assistance and warning prototype for recognizing airport surface types and horizontal markings in video recorded at Poznan Airport. Two manually annotated segmentation datasets were prepared from GoPro HERO8 video acquired from a vehicle perspective: a four-class surface dataset covering asphalt, concrete, paving blocks and grass, and a three-class marking dataset covering red, white and yellow lines. The study compares You Only Look Once (YOLO) variants YOLOv8 and YOLOv11 with U-Net, DeepLabV3 and SegFormer under a common 512-by-512 input resolution and evaluates both model-level quality and complete video-application behavior. For semantic segmentation, SegFormer achieved the highest validation results, with Intersection over Union (IoU)/Dice of 0.7657/0.8624 for surfaces and 0.8852/0.9380 for markings. Among YOLO models, YOLOv8m obtained the highest surface mean average precision at an IoU threshold of 0.50 (mAP@50) of 0.7847, whereas YOLOv8s obtained the highest marking mAP@50 of 0.8449. On video recordings, paired YOLO configurations processed approximately 15–16 frames per second (FPS) on a personal computer (PC), while U-Net, DeepLabV3 and SegFormer processed approximately 10–11 FPS. A YOLOv8n pair compiled for Raspberry Pi 5 with Raspberry Pi AI HAT+ Hailo-8 reached 10.05 detection FPS and 18.15 processing FPS without GUI rendering. Under the adopted evaluation protocol, SegFormer achieved the highest mask quality in the conducted comparison, while the paired YOLOv8n configuration demonstrated embedded throughput feasibility. Full article

Accurate cloud segmentation from ground-based fisheye camera imagery is essential for solar irradiance forecasting and photovoltaic system optimization. Traditional computer vision approaches, such as HSV thresholding and K-means clustering, face significant limitations when applied globally to sky images due to the spectral similarity between cloud regions and sky areas under varying atmospheric conditions. This study presents a hybrid methodology that leverages YOLOv8 instance segmentation to provide contextual cloud regions followed by refined HSV thresholding within these detected areas. The approach incorporates solar trajectory modeling using pvlib for accurate sun disk detection and exclusion, preventing false cloud classification. The methodology was developed and validated at the CIESOL using Mobotix Q71 fisheye cameras, and later tested in Antofagasta (Chile) and Huelva (Spain). The YOLOv8l-seg model achieved a mask precision of 0.821 and box mAP@0.5 of 0.680 on validation data. The results show a promising correlation with radiometric measurements such as clearness index $k_t$ and diffuse fraction $k_d$ in preliminary validation cases. While YOLOv8 demonstrates good cross-site generalization, HSV thresholding requires camera-specific calibration for optimal performance. The method addresses the context-dependency limitations of traditional algorithms, though computational performance and broader validation remain areas for future work. Full article

Search-and-rescue (SAR) missions following earthquakes and other disasters require aerial video perception systems that do more than detect objects in isolated frames. Operators must maintain the identities of access points, vehicles, responders, hazards, and other mission-relevant targets despite UAV ego-motion, dust, debris, occlusion, scale variation, and abrupt scene transitions. This paper presents E’CHIT (Edge-Oriented Colour Histogram Instance-Guided Tracking), a deployment-oriented, operator-centric UAV tracking framework for real-world disaster-response applications. Its primary scientific contribution is an identity-stabilised, detector-assisted tracking methodology. YOLOv8-seg proposals trained on D’RespNeT initialise and refresh tracks; a Custom-RE3 recurrent module propagates target states through short detector dropouts; and a lightweight EOMC verifier, based on edge orientation, mean colour, and shape consistency, determines whether tracks should be accepted, refreshed, or reacquired. A scene-cut watchdog that combines luminance mean absolute difference (MAD) with HSV histogram divergence prevents stale identities from carrying over after hard edits or sudden feed changes. Custom-RE3 is the continuation module implemented and evaluated in this study. The surrounding E’CHIT wrapper follows an initialise–reseed–verify–reset cycle and is tracker-adaptable at the software-interface level: another compatible SOT or MOT continuation module can be integrated through adapter modifications, state and bounding-box conversion, and method-specific retuning, followed by independent validation. All reported quantitative results therefore apply to the Custom-RE3 implementation. D’RespNeT, the optional reinforcement learning (RL) warm start, the HUD, and the deployment stack support this central tracking contribution. D’RespNeT provides 28 polygon-annotated SAR classes. An author-developed PPO/SAC script is used only during offline detector training. In the reported runs, it produces different early optimisation trajectories for selected difficult or under-represented classes, while the default supervised schedule remains the strongest final global mAP reference. No RL policy runs during deployment; the detector architecture, parameter count, and inference graph remain unchanged. Evaluation on D’RespNeT and authentic disaster-response UAV footage shows that E’CHIT increases Success@IoU ≥ 0.5 from 0.62 to 0.79, reduces identity switches by approximately 71%, and maintains real-time 1080p performance, achieving 164–330 FPS for single-target tracking and 24–100+ FPS for end-to-end multi-target operation on an RTX-class GPU using FP16. The VOT2014, NT-VOT211, and VOTS2024 figures reproduce historical result spaces reported in the literature and include a clearly labelled, non-official E’CHIT operating-point marker solely for context. This marker was not produced using the corresponding official datasets, toolkits, reset rules, or submission routes; it is excluded from the primary quantitative claims and must not be interpreted as a leaderboard rank or a protocol-identical comparison. Overall, the system demonstrates how identity-stable UAV tracks can provide actionable operator cues for target monitoring, entry-point assessment, and UAV–UGV/ground-team coordination in cluttered disaster scenes. Full article

Shrimp farming is a strategic productive activity for Ecuador; however, pond inspection still substantially depends on manual observation and fragmented visual records. This study describes a prototype mobile/web architecture for image capture, storage, and result visualization and, as a separate experiment, evaluates supervised multiclass instance segmentation on public proxy data. A unified dataset of 4508 images was constructed from three external sources using the classes foam, floater, and shrimp. YOLOv8s-seg was used as the internal reference baseline and YOLOv11s-seg as the comparison candidate; both were trained under the same configuration and evaluated on 445 test images. YOLOv8s-seg achieved mAP50 values of 0.678 for BOX and 0.621 for MASK, whereas YOLOv11s-seg achieved 0.677 and 0.616, respectively. Their isolated GPU inference times were 10.4 and 10.2 ms/img. A weighted global experimental performance index comprising 90% predictive quality and 10% inference efficiency reached 0.596 and 0.593, respectively. Both models performed strongly for floater and shrimp, whereas foam showed low recall and zero MASK mAP50 because of the small number of positive test images and heterogeneous annotations. The evaluated task is supervised segmentation of proxy visual categories rather than anomaly detection in its conventional methodological sense, and no general improvement or practical superiority of one architecture was demonstrated. The study does not calibrate an operational confidence threshold or alert-persistence rule, implement adaptive or online learning, validate images from Ecuadorian production ponds, deploy the unified detector within the API, or demonstrate end-to-end real-time monitoring. Consequently, the results constitute a controlled experimental baseline and must not be interpreted as evidence of operational performance or local-domain generalization. Full article

In-flight wingtip docking can connect multiple UAVs into a high-aspect-ratio chained-wing configuration, offering potential improvements in aerodynamic efficiency, endurance, and cruise performance. Reliable close-range 6-DoF relative pose estimation is essential for precise docking; however, existing studies have focused primarily on aerodynamic characteristics, docking mechanisms, and guidance and control, while robust monocular pose estimation under partial occlusion and image degradation remains insufficiently investigated. To address this gap, a cooperative-target-based monocular vision method is proposed for six-degree-of-freedom relative pose estimation during in-flight wingtip docking in a chained-wing UAV system. An asymmetric seven-ring planar cooperative target is designed to reduce feature-identification ambiguity and retain sufficient geometric constraints under partial occlusion. The front end combines YOLOv8n-seg instance segmentation with local inner–outer ring refinement and topology-based feature classification. According to target visibility, the back end adaptively selects seven-, five-, or four-point pose-estimation modes, refines valid solutions by minimizing reprojection error, and removes isolated suspicious candidates when necessary. The method is evaluated on a controlled indoor hardware-in-the-loop platform as a pre-flight assessment of its incremental measurement performance. In controlled incremental-response experiments, the mean absolute adjacent-increment errors do not exceed $0.08\mathrm{mm}$ in translation and $0.13°$ in rotation. The complete method achieves a pose-solving success rate of 99.07%. Experiments involving progressive wing occlusion and synthetic directional motion blur further demonstrate that the method can provide stable and continuous relative pose output under challenging observation conditions. Full article

Background/Objectives: Third molar development is one of the most frequently used dental indicators in forensic age assessment because its maturation continues through adolescence and early adulthood. Manual staging on panoramic radiographs requires experience and may be affected by observer variability, especially in transitional developmental stages. This study evaluated the performance of segmentation-based deep learning models for automatic third molar localization and Demirjian-based developmental stage classification on panoramic radiographs. Methods: The study used a two-stage segmentation framework. In the first stage, third molar localization was performed using 737 panoramic radiographs and 2736 annotations for teeth 18, 28, 38, and 48. In the second stage, developmental stage classification was performed using 695 panoramic radiographs and 2533 annotations grouped as AB, CD, EF, and GH according to Demirjian developmental stages. A supplementary 18-year threshold segmentation analysis was added using 695 panoramic radiographs and 2573 labels divided into training, validation, and test sets at an 80:10:10 ratio. YOLO-based segmentation models were trained using AdamW optimization and learning-rate settings recorded in the training logs. Model performance was evaluated using TP, FP, FN, precision, recall, F1-score, accuracy, mAP, Dice coefficient, Jaccard index, and performance curves. For instance segmentation, true negatives were not calculated because the number of background non-objects is not finite or clinically meaningful. Results: The third molar localization model yielded TP = 256, FP = 4, and FN = 11 on the test set. The overall accuracy was 0.9446, mAP@0.5 was 0.980, mAP@0.5:0.95 was 0.875, Dice coefficient was 0.9064, and Jaccard index was 0.8696. In developmental stage classification, yolo11x-seg produced the most balanced segmentation profile, with an accuracy of 0.7272, Dice coefficient of 0.5783, and Jaccard index of 0.5529. In the supplementary 18-year threshold segmentation analysis, yolov8x-seg produced TP = 172, FP = 37, and FN = 32. Overall accuracy was 0.8152, precision was 0.8230, recall was 0.8431, F1-score was 0.8329, mAP@0.5 was 0.635, and mAP@0.5:0.95 was 0.563. Class-wise results showed better performance for the under-18 class than for the 18-years-and-older class. Conclusions: YOLO-based segmentation models can localize third molars on panoramic radiographs with high performance. Developmental stage classification is more difficult than anatomical localization because the radiographic boundaries between adjacent developmental stages are gradual rather than discrete. The added 18-year threshold analysis provides a clinically relevant age threshold experiment, but it also shows that class imbalance and weak segmentation of the 18-years-and-older class limit direct forensic use. Segmentation-based third molar analysis should therefore be interpreted as a visually auditable decision-support workflow rather than a stand-alone legal age determination tool. Full article

Detector-assisted promptable segmentation can reuse box annotations, but prompt design creates an accuracy–efficiency trade-off in dense scenes. We study a training-free interface in which a detector trained with bounding-box annotations localizes stacked cartons and a frozen segmenter generates masks. Four interfaces are evaluated: Point-Single, Box, ambiguity-aware Point-Max, and Grid-Iter with sparse positive points and one mask-logit feedback pass. On a leakage-audited split, COCO mask AP, fixed-threshold Hungarian-matched metrics, paired bootstrap uncertainty, and synchronized latency were measured. On 773 test images, Box and Grid-Iter achieved mask AP values of 0.8984 and 0.8988. Their paired AP difference was 0.0007 (95% CI, $-0.0035$ to 0.0047), while matched-mIoU and F1 intervals also included zero. Box required 604.67 ms/image versus 677.33 ms/image for Grid-Iter. A YOLO26s–SAM 3 Box reference achieved a mask AP value of 0.9115 at 320.74 ms/image, whereas mask-supervised YOLOv9c-seg achieved a value of 0.9235 AP on the same cleaned split. The results indicate a metric-dependent Pareto trade-off rather than overall Grid-Iter superiority and position detector-to-segmenter prompting as a reproducible annotation–accuracy option for densely stacked cartons. Full article

Glacial lakes are sensitive indicators of climate-driven cryospheric change, and their accurate mapping provides fundamental spatial information for water-resource assessment and glacial lake outburst flood (GLOF) hazard assessment. In southeastern Tibet, automatic extraction remains difficult because glacial lakes are small and easily confused with snow, mountain shadows, dark bedrock, riverine wetlands, and non-glacial water bodies. In this study, we integrate optical bands and water indices from Sentinel-2, topographic information derived from a digital elevation model, and radar backscatter from Sentinel-1 into a nine-channel multimodal dataset, and develop an improved YOLO11-seg model that combines spatial-to-depth downsampling, multi-scale attention, long-range context modeling, and content-aware upsampling to enhance small-lake detection, background suppression, and boundary delineation. Compared with U-Net, DeepLabV3+, YOLOv8-seg, YOLO11-seg, YOLO12-seg, and YOLO26-seg, the proposed model achieved the highest F1-Score of 0.9205 and an mAP50(M) of 0.9331, while its F1-Score and IoU on the independent test set reached 0.9209 and 0.8533, respectively. Using remote sensing imagery acquired in 2024, the model extracted 4766 glacial lakes in southeastern Tibet, covering 428.13 km²; 80.84% of these lakes were smaller than 0.10 km². The results demonstrate an effective and reproducible framework for automatic glacial lake mapping in complex alpine environments. Full article

With the expansion of urban underground space, the structural safety of underground infrastructure has become increasingly critical. The urban underground utility tunnel is a typical deeply buried lifeline project. Its internal environment is humid and confined, and the structure is subjected to long-term heavy loads and earth pressure. Under such conditions, micro-cracks readily propagate into leakage channels and eventually cause structural damage. However, the slender morphology, low contrast, and complex background of cracks make it difficult for traditional inspection methods and general-purpose models to achieve reliable identification and accurate segmentation. This study proposes a two-stage framework that combines YOLOv8 with SAM2 and incorporates a coordinate attention module for high-quality crack segmentation. For image processing, a topology-aware post-processing strategy is introduced, together with a scoring function based on crack morphological features and a post-processing constraint mechanism, to ensure crack continuity and geometric consistency and to mitigate the over-segmentation that may occur during segmentation. YOLOv8-SAM2 achieves 85.2%, 90.5%, 77.2%, and 71.6% in mIoU, mDice, Recall, and Precision, respectively. Compared with YOLOv8-seg, mIoU and Precision are improved by 46.3 and 17.2 percentage points, respectively; compared with the baseline SAM2, mIoU and Precision are improved by 23.0 and 16.9 percentage points, respectively. In summary, on the self-built underground utility tunnel crack dataset, the proposed model significantly outperforms standalone YOLOv8 segmentation and the direct application of SAM2 in terms of intersection-over-union and precision, demonstrating its high-quality segmentation capability. Full article

Visible fracture detection in underground roadways is challenging because fracture targets are often elongated, weakly contrasted, irregularly distributed, and easily confused with complex rock-wall textures. In addition, uneven illumination, dust interference, and blurred boundaries further reduce the reliability of conventional crack detection and segmentation methods. To improve fracture instance segmentation under such conditions, this study proposes YOLOv13n-seg-crack, an improved lightweight instance segmentation model based on a self-constructed YOLOv13n-seg baseline. The proposed model introduces three main improvements. First, a C2f-CA module is embedded into the backbone to enhance spatial-position perception and directional feature representation for elongated fractures. Second, a shallow high-resolution branch and auxiliary feature paths, denoted as B2 + H2 + P2, are constructed to strengthen the transmission of fine edge and texture information for small and discontinuous fracture targets. Third, an Edge-aware SIoU (EA-SIoU) loss is designed by adding edge-consistency and aspect-ratio constraints, thereby improving bounding-box localization for narrow and irregular fracture regions. Experiments were conducted on the public Crack Segmentation Dataset and an expanded self-built underground roadway dataset collected at the Woniushan Experimental Base. On the public dataset, YOLOv13n-seg-crack achieved detection Precision, Recall, mAP50, and mAP50:95 of 84.56%, 65.49%, 71.51%, and 52.72%, respectively, and mask Precision, Recall, mAP50, and mAP50:95 of 74.94%, 60.38%, 59.52%, and 21.99%, respectively. Compared with YOLOv13n-seg, the detection mAP50 and mask mAP50 increased by 1.91 and 3.19 percentage points, respectively, while the model maintained an inference speed of 168.73 FPS. Repeated-seed experiments, ablation studies, and degraded-image tests further demonstrate the stability and robustness of the proposed improvements. On the self-built underground roadway dataset containing 100 images and 118 annotated fracture instances, YOLOv13n-seg-crack improved detection mAP50 from 68.72% to 73.36% and mask mAP50 from 30.76% to 33.74%. These results indicate that the proposed method provides an effective and lightweight solution for visible fracture detection and instance segmentation in complex underground roadway scenes. Full article

Rock fissure parameters, such as length, width, and density, are essential for analyzing the progressive instability of open-pit mine slopes. Under the combined effects of engineering disturbance, geological conditions, and environmental factors, slope fissures continuously propagate and evolve. However, large variations in fissure scale, complex rock-surface textures, blurred boundaries, and weak micro-fissure features increase the difficulty of intelligent fissure segmentation, identification, and parameter extraction. Consequently, many mining enterprises still rely on manual interpretation, which is time-consuming and susceptible to subjective errors. To address these challenges, this study develops Crack-YOLO, a task-oriented fissure detection and instance-segmentation model based on YOLOv8-Seg. A total of 500 original UAV images were collected from multiple open-pit mines and processed to construct a dataset containing 3600 fissure image patches, including 3240 images for training and 360 images for testing. In Crack-YOLO, selected C2f modules are replaced with contextual semantic enhancement modules (CoT Blocks), and a texture information enhancement module (SM Block) is incorporated to strengthen contextual semantic representation and fine-grained texture-feature extraction. The model achieved segmentation precision, recall, mAP50, and mAP50:95 values of 0.896, 0.787, 0.854, and 0.392, respectively. For object detection, the corresponding values were 0.968, 0.862, 0.959, and 0.773, respectively. The segmentation results were further processed using K3M skeleton extraction and physical-scale calibration to quantitatively extract geometric parameters, including fissure length, equivalent average width, and azimuth. Validation using an image containing seven representative fissures yielded mean absolute errors of 0.016 m, 0.010 m, and 0.90° for fissure length, equivalent average width, and azimuth, respectively, indicating the feasibility of the proposed parameter-quantification workflow. In an application test conducted in a typical open-pit mine scene, the proposed workflow identified 196 fissures within approximately 22 s and quantitatively analyzed their geometric parameters and distribution characteristics. The results indicate that the proposed method has potential for fissure identification and geometric-parameter quantification in open-pit mine slopes and may provide quantitative data support for slope-fissure monitoring and stability analysis. Full article

To address inaccurate weed segmentation under crop overlap, occlusion, and complex field backgrounds, this study developed a combined method integrating DViT-YOLOv8-seg with confidence-guided SLIC voting. The dataset contained 1872 field images (800 × 600 pixels) of Guangzhou soft-stem lettuce and four common weed species: Eleusine indica, Digitaria sanguinalis, Portulaca oleracea, and Amaranthus blitum. All weed species were merged into one weed class, while lettuce, soil, and other field regions were treated as non-weed. Real-ESRGAN and data augmentation enhanced the training samples; Dual-ViT strengthened global–local feature interaction; GSConv reduced redundant computation; BiFPN improved multi-scale fusion; and SLIC refined ambiguous boundaries. After super-resolution preprocessing and three-fold expansion, baseline mPA increased by 10.9 percentage points. The improved network achieved 88.3% mPA at 7.9 GFLOPs, corresponding to +3.6 percentage points and -1.0 GFLOPs relative to the baseline. SLIC voting increased FWIoU to 95.6%, 4.1 percentage points above the network without SLIC. Compared with YOLOv5-seg and Fast-SCNN, mPA improved by 2.0 and 6.5 percentage points, respectively; GFLOPs were 87.7% and 95.5% lower than those of YOLOv5-seg and DeepLabv3+, respectively. The method therefore provides a favorable trade-off between segmentation accuracy and theoretical network computation for complex lettuce field imagery. Full article

This paper presents an integrated framework for automated detection of surface defects in infrastructures using unmanned aerial vehicles (UAVs), comprising 3D model-based adaptive path planning, high-resolution image acquisition, and an attention-enhanced instance segmentation model. However, existing approaches face two key limitations: (i) conventional UAV path planning lacks adaptive trajectory correction for non-horizontal bridge geometries, and (ii) instance segmentation models for infrastructure defects have not been systematically optimized for both accuracy and edge-device deployability. To address these gaps, the proposed framework was trained on 3625 annotated images covering two defect categories (spalling and cracking) and preliminarily validated through a proof-of-concept field study on a concrete viaduct section where seven spalling and one reinforcement exposure were detected. The methodology consists of three core components: (i) 3D model-based adaptive path planning, (ii) high-resolution image acquisition under variable infrastructure geometries, and (iii) an improved instance segmentation model based on YOLOv8-seg. To ensure consistent imaging geometry, a segmented linear interpolation method is introduced to adaptively correct flight trajectories for non-horizontal infrastructure sections. For damage detection, we propose a structurally enhanced YOLOv8-seg model, denoted as YOLOv8-seg-ECAC2f-all, which integrates Efficient Channel Attention (ECA) modules into a fully modified backbone architecture. Compared to the baseline YOLOv8-seg, the proposed model achieves a mean average precision (mAP_50–95) of 91.8% for bounding box detection and 61.7% for instance segmentation, corresponding to improvements of 7.7% and 2.8%, respectively. The framework was preliminarily validated on a concrete viaduct of the Guangfojiangzhu Expressway, achieving 100% inspection coverage and detection of all eight ground-truth surface defects (seven spalling and one reinforcement exposure) in this pilot study, including two minor spalling cases (≤0.5 m²) missed by manual inspection. These results demonstrate the technical feasibility of the proposed framework for real-world concrete bridge inspection and its potential for reducing manual inspection effort while improving detection sensitivity for minor defects (${\mathrm{m}\mathrm{A}\mathrm{P}}_{50–95}$). Full article

Efficient conservation prioritization of Huizhou historic buildings requires joint consideration of visible deterioration and value-bearing architectural elements. This study developed a deep learning-based decision-support framework using an original field dataset of 500 facade images and 2197 audited annotations from five towns in Shexian County, China. Separate YOLOv11n and YOLOv11n-seg models detected six decorative or typological element classes and segmented cracks, spalling, and stains, respectively. Model outputs were converted into a pathology severity index (PSI) based on mask-pixel-area ratios and a decorative value index (DVI) based on weighted element counts; the two indices were combined into a conservation priority index (CPI), whose weighting was examined through sensitivity analysis. On the validation set, decorative-element detection yielded a mAP@50 of 0.655, and pathology segmentation yielded a mask mAP@50 of 0.586. In a preliminary application to four held-out buildings, model-derived priority categories matched the blind ratings of three conservation experts in three cases. The framework offers interpretable evidence for preliminary screening and resource-allocation discussions, but it does not replace field diagnosis. Larger balanced datasets, external building-level validation, and metric calibration are required before regional deployment. Full article