Accurate retinal vessel segmentation supports quantitative vascular analysis in assessing ocular and systemic diseases. Yet, its clinical scalability is constrained by limited pixel-level annotations and domain shift across heterogeneous fundus datasets. Thus, this study proposes a domain-adaptive retinal vessel segmentation model (DA-VesselNet), a weakly supervised approach that transfers vessel-segmentation knowledge from annotated source datasets to the unannotated Retinal Fundus Multi-Disease Image Dataset (RFMiD). The model was trained on DRIVE, CHASE_DB1, and FIVES, and adapted to RFMiD. The ResNet50 encoder with an attention-gated U-Net decoder, confidence-aware pseudo-label supervision, and perturbation consistency regularisation were used for the training process. Results on the held-out CHASE_DB1 indicated that DA-VesselNet achieved a Dice score of 0.5778, an Intersection over Union (IoU) of 0.4082, and an Area Under the Curve (AUC) of 0.9518. On 200 held-out FIVES test images with different pathological features, it achieved a Dice score of 0.7447 and an AUC of 0.9733, outperforming the source-only baseline model. To assess adaptation independently of source-adjacent data, the model was further tested on STARE and HRF, two domains excluded entirely from source training. Adaptation improved Dice by 0.0469 on STARE and 0.0056 on HRF with AUC gains of 0.0140 and 0.0121, respectively. Ablation analysis identified source-anchored supervision as the dominant contributor to performance. The adapted model was subsequently applied to generate vessel pseudo-labels for the RFMiD target domain, providing a structural resource for future vessel-informed analysis. These findings demonstrate that DA-VesselNet offers a scalable solution for creating clinically relevant pseudo-labelled vessels in fundus imaging with limited annotations.
M. O. Oladele, O. A. Alimi, O. Olugbara· Applied Sciences· 0 citations
Ensuring reliable equipment operations are critical for production efficiency and safety compliance in manufacturing industries. Unexpected machine breakdowns not only disrupt operations but also increase maintenance costs and safety risk. Traditional approaches — whether reactive or preventive — often fail to incorporate real-time equipment data and often overlook early fault indicators. Predictive maintenance strategies based on artificial intelligence models, address these shortcomings by monitoring machine conditions in real time, detect anomalies and forecast failures. Thus, this study proposes a refined deep neural network to classify machine health and estimate failure probability, thereby emphasizing Remaining Useful Life (RUL) prediction as a strategy for optimizing maintenance scheduling. In this study, the AI4I 2020 Predictive Maintenance dataset, which contains 10,000 records of machine operating conditions, including air and process temperatures, rotational speed, torque, tool wear, product type, and failure status was used for the evaluation of the proposed model. To improve the proposed model's performance, detailed preprocessing steps were deployed on the dataset. These steps include some preliminary categorical encoding and feature standardization on the dataset while class weighting, SMOTE and focal loss were deployed for handling class imbalance issues. According to the results achieved, the proposed model performed better across all metrics considered in comparison with similar models including baseline models. By integrating data-driven AI techniques with predictive maintenance strategies, this research study demonstrates how manufacturing plants can reduce downtime, extend machine lifespan, and minimize unnecessary maintenance interventions.
O. A. Alimi, O.C. Tshidavhu, N. Khanyile et al.· 2026 6th International Confe...· 0 citations