Real-time visual SLAM based YOLO-Fastest for dynamic scenes

Measurement Science and Technology, Journal Year: 2024, Volume and Issue: 35(5), P. 056305 - 056305

Published: Feb. 5, 2024

Abstract Within the realm of autonomous robotic navigation, simultaneous localization and mapping (SLAM) serves as a critical perception technology, drawing heightened attention in contemporary research. The traditional SLAM systems perform well static environments, but real physical world, dynamic objects can destroy geometric constraints system, further limiting its practical application world. In this paper, robust RGB-D system is proposed to expand number points scene by combining with YOLO-Fastest ensure effectiveness model construction, then based on that, new thresholding designed differentiate features objection bounding box, which takes advantage double polyline residuals after reprojection filter feature points. addition, two Gaussian models are constructed segment moving box depth image achieve effect similar instance segmentation under premise ensuring computational speed. experiments conducted sequences provided TUM dataset evaluate performance method, results show that root mean squared error metric absolute trajectory algorithm paper has at least 80% improvement compared ORB-SLAM2. Higher robustness environments both high low DS-SLAM Dynaslam, effectively provide intelligent navigation for mobile robots.

Language: Английский

---

Artificial-Intelligence-Enhanced Analysis of In Vivo Confocal Microscopy in Corneal Diseases: A Review

Diagnostics, Journal Year: 2024, Volume and Issue: 14(7), P. 694 - 694

Published: March 26, 2024

Artificial intelligence (AI) has seen significant progress in medical diagnostics, particularly image and video analysis. This review focuses on the application of AI analyzing vivo confocal microscopy (IVCM) images for corneal diseases. The cornea, as an exposed delicate part body, necessitates precise diagnoses various conditions. Convolutional neural networks (CNNs), a key component deep learning, are powerful tool data highlights applications diagnosing keratitis, dry eye disease, diabetic neuropathy. It discusses potential detecting infectious agents, nerve morphology, identifying subtle changes fiber characteristics However, challenges still remain, including limited datasets, overfitting, low-quality images, unrepresentative training datasets. explores augmentation techniques importance feature engineering to address these challenges. Despite made, present, such “black-box” nature models need explainable (XAI). Expanding fostering collaborative efforts, developing user-friendly tools crucial enhancing acceptance integration into clinical practice.

Language: Английский

---

Solar Cell Defects Detection Based on Photoluminescence Images and Upgraded YOLOv5 Model

Journal of Engineering, Journal Year: 2025, Volume and Issue: 2025(1)

Published: Jan. 1, 2025

Solar cells (SCs) are prone to various defects, which affect energy conversion efficiency and even cause fatal damage photovoltaic modules. In this paper, photoluminescence (PL) imaging is used visualize SC based on a detection method the YOLOv5 model explored. At same time, five data enhancement methods such as Mosaic, Mixup, HSV transformation, Gaussian noise, rotation transformation introduced improve representativeness of set enhance ability model. Second, C2f module designed network model’s fuse features. order further convolutional network’s capture target features, series SPPF combined with soft pooling proposed reduce number repeated operations, efficiency, focus extracting higher level features from input. Experimental results show that optimized mAP reaches 91.5%, 20.3% than original The increase some defect types 50.4%, speed 24.2 FPS. capability for has been significantly enhanced, meeting requirements at time.

Language: Английский

---

RFCS-YOLO: Target Detection Algorithm in Adverse Weather Conditions via Receptive Field Enhancement and Cross-Scale Fusion

Sensors, Journal Year: 2025, Volume and Issue: 25(3), P. 912 - 912

Published: Feb. 3, 2025

The paper proposes a model based on receptive field enhancement and cross-scale fusion (RFCS-YOLO). It addresses challenges like complex backgrounds problems of missing mis-detecting traffic targets in bad weather. First, an efficient feature extraction module (EFEM) is created. reconfigures the backbone network. This helps to make better improves its ability extract features at different scales. Next, (CSF) introduced. uses coordinate attention mechanism (RFCA) fuse information from scales well. also filters out noise background that might interfere. Also, new Focaler-Minimum Point Distance Intersection over Union (F-MPDIoU) loss function proposed. makes converge faster deals with issues leakage false detection. Experiments were conducted expanded Vehicle Detection Adverse Weather Nature dataset (DWAN). results show significant improvements compared conventional You Only Look Once v7 (YOLOv7) model. mean Average Precision ([email protected]), precision, recall are enhanced by 4.2%, 8.3%, 1.4%, respectively. 86.5%. frame rate 68 frames per second (FPS), which meets requirements for real-time A generalization experiment was using autonomous driving SODA10M. [email protected] achieved 56.7%, 3.6% improvement original result demonstrates good proposed method.

Language: Английский

---

Recognizing Adenoid Hypertrophy from Facial Images with Multi-scale Feature Fused State Space Model

Lecture notes in computer science, Journal Year: 2025, Volume and Issue: unknown, P. 228 - 237

Published: Jan. 1, 2025

Language: Английский

---