GT-YOLO: Nearshore Infrared Ship Detection Based on Infrared Images

Journal of Marine Science and Engineering, Journal Year: 2024, Volume and Issue: 12(2), P. 213 - 213

Published: Jan. 25, 2024

Traditional visible light target detection is usually applied in scenes with good visibility, while the advantage of infrared that it can detect targets at nighttime and harsh weather, thus being able to be ship complex sea conditions all day long. However, coastal areas where density ships high there a significant difference scale, this lead missed some dense small targets. To address issue, paper proposes an improved model based on YOLOv5s. Firstly, article designs feature fusion module attention mechanism enhance network introduces SPD-Conv improve accuracy low-resolution images. Secondly, by introducing Soft-NMS, also addressing issue detections occlusion situations. Finally, algorithm increased mAP0.5 1%, mAP0.75 5.7%, mAP0.5:0.95 5% dataset. A large number comparative experiments have shown effective improving capabilities.

Language: Английский

---

Panicle-Cloud: An Open and AI-Powered Cloud Computing Platform for Quantifying Rice Panicles from Drone-Collected Imagery to Enable the Classification of Yield Production in Rice

Plant Phenomics, Journal Year: 2023, Volume and Issue: 5

Published: Jan. 1, 2023

Rice ( Oryza sativa ) is an essential stable food for many rice consumption nations in the world and, thus, importance to improve its yield production under global climate changes. To evaluate different varieties’ performance, key yield-related traits such as panicle number per unit area (PNpM 2 are indicators, which have attracted much attention by plant research groups. Nevertheless, it still challenging conduct large-scale screening of panicles quantify PNpM trait due complex field conditions, a large variation cultivars, and their morphological features. Here, we present Panicle-Cloud, open artificial intelligence (AI)-powered cloud computing platform that capable quantifying from drone-collected imagery. facilitate development AI-powered detection models, first established diverse dataset was annotated group specialists; then, integrated several state-of-the-art deep learning models (including preferred model called Panicle-AI) into Panicle-Cloud platform, so nonexpert users could select pretrained detect own aerial images. We trialed AI with images collected at attitudes growth stages, through right timing image resolutions phenotyping were identified. Then, applied 2-season breeding trial valid biological relevance classified using platform-derived hundreds varieties. Through correlation analysis between computational manual scoring, found reliably, based on high accuracy. Hence, trust our work demonstrates valuable advance rice, provides useful toolkit enable breeders screen desired varieties conditions.

Language: Английский

---

BDC-YOLOv5: a helmet detection model employs improved YOLOv5

Signal Image and Video Processing, Journal Year: 2023, Volume and Issue: 17(8), P. 4435 - 4445

Published: Aug. 14, 2023

Language: Английский

---

A Lightweight Radar Ship Detection Framework with Hybrid Attentions

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(11), P. 2743 - 2743

Published: May 25, 2023

One of the current research areas in synthetic aperture radar (SAR) processing fields is deep learning-based ship detection SAR imagery. Recently, images has achieved continuous breakthroughs precision. However, determining how to strike a better balance between precision and complexity algorithm very meaningful for real-time object real application scenarios, attracted extensive attention from scholars. In this paper, lightweight framework named multiple hybrid attentions detector (MHASD) with mechanisms proposed. It aims reduce without loss First, considering that features are not inconspicuous compared other images, residual module (HARM) developed deep-level layer obtain rapidly effectively via local channel parallel self-attentions. Meanwhile, it also capable ensuring high model. Second, an attention-based feature fusion scheme (AFFS) proposed model neck further heighten object. AFFS constructs develops fresh (HAFFM) upon spatial guarantee applicability The Large-Scale Ship Detection Dataset-v1.0 (LS-SSDD-v1.0) experimental results demonstrate MHASD can speed (improving average by 1.2% achieving 13.7 GFLOPS). More importantly, experiments on Dataset (SSDD) method less affected background such as ports rocks.

Language: Английский

---

SE-Lightweight YOLO: Higher Accuracy in YOLO Detection for Vehicle Inspection

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(24), P. 13052 - 13052

Published: Dec. 7, 2023

Against the backdrop of ongoing urbanization, issues such as traffic congestion and accidents are assuming heightened prominence, necessitating urgent practical interventions to enhance efficiency safety transportation systems. A paramount challenge lies in realizing real-time vehicle monitoring, flow management, control within infrastructure mitigate congestion, optimize road utilization, curb accidents. In response this challenge, present study leverages advanced computer vision technology for detection tracking, employing deep learning algorithms. The resultant recognition outcomes provide management domain with actionable insights optimizing signal light through data analysis. demonstrates applicability SE-Lightweight YOLO algorithm, presented herein, showcasing a noteworthy 95.7% accuracy recognition. As prospective trajectory, research stands poised serve pivotal reference urban laying groundwork more efficient, secure, streamlined system future. To solve existing problems type recognition, need be improved, alongside resolving slow speed, others. paper, we made innovative changes based on YOLOv7 framework: added SE attention transfer mechanism backbone module, model achieved better results, 1.2% improvement compared original YOLOv7. Meanwhile, replaced SPPCSPC module SPPFCSPC which enhanced trait extraction model. After that, applied field monitoring. This can assist transportation-related personnel monitoring aid creating big transportation. Therefore, has good application prospect.

Language: Английский

---

Unmanned Ship Identification Based on Improved YOLOv8s Algorithm

Computers, materials & continua/Computers, materials & continua (Print), Journal Year: 2024, Volume and Issue: 78(3), P. 3071 - 3088

Published: Jan. 1, 2024

Aiming at defects such as low contrast in infrared ship images, uneven distribution of size, and lack texture details, which will lead to unmanned leakage misdetection slow detection, this paper proposes an detection model based on the improved YOLOv8 algorithm (R_YOLO).The incorporates Efficient Multi-Scale Attention mechanism (EMA), efficient Reparameterized Generalized-feature extraction module (CSPStage), small target header, Repulsion Loss function, context aggregation block (CABlock), are designed improve model's ability detect targets multiple scales speed inference.The is validated detail two vessel datasets.The comprehensive experimental results demonstrate that, dataset, YOLOv8s exhibits improvements various performance metrics.Specifically, compared baseline algorithm, there a 3.1% increase mean average precision threshold 0.5 (mAP (0.5)), 5.4% recall rate, 2.2% mAP (0.5:0.95).Simultaneously, while less than 5 times parameters, (0.5) frames per second (FPS) exhibit 1.7% more 3 times, respectively, CAA_YOLO algorithm.Finally, evaluation indexes visible light data set have shown improvement 4.5%.

Language: Английский

---

Light-YOLO: a lightweight detection algorithm based on multi-scale feature enhancement for infrared small ship target

Complex & Intelligent Systems, Journal Year: 2025, Volume and Issue: 11(2)

Published: Jan. 7, 2025

Infrared-based detection of small targets on ships is crucial for ensuring navigation safety and effective maritime traffic management. However, existing ship target models often encounter missed detections struggle to achieve both high accuracy real-time performance at the same time. Addressing these challenges, this study presents Light-YOLO, a lightweight model detection. Within YOLOv8 network architecture, Light-YOLO replaces conventional convolutions with snake convolutions, effectively addressing issue inadequate point receptive fields targets, thereby enhancing their Additionally, Multi-Scale Feature Enhancement Module (MFEB) introduced refine focus low-level features through multi-scale selection strategies, mitigating issues such as interference from image backgrounds noise during Furthermore, novel loss function designed dynamically adjust proportions its components training, improving regression towards real annotation boxes localization ability boxes. Experimental results demonstrate that outperforms YOLOv8n, achieving optimal an infrared dataset 9.2G FLOPs. It notably enhances accuracy, recall rate, average precision by 1.76%, 0.83%, 2.27%, respectively.

Language: Английский

---

RTR_Lite_MobileNetV2: A Lightweight and Efficient Model for Plant Disease Detection and Classification

Current Plant Biology, Journal Year: 2025, Volume and Issue: unknown, P. 100459 - 100459

Published: Feb. 1, 2025

Language: Английский

---

Ship_YOLO: General ship detection based on mixed distillation and dynamic task-aligned detection head

Ocean Engineering, Journal Year: 2025, Volume and Issue: 323, P. 120616 - 120616

Published: Feb. 17, 2025

Language: Английский

---

An improved YOLOv5 model for aeolian saltating particle recognition in high-speed video

Measurement, Journal Year: 2025, Volume and Issue: unknown, P. 117321 - 117321

Published: March 1, 2025

Language: Английский

---