Introduction: Precise semantic segmentation of microbial alterations is paramount for their evaluation and treatment. This study focuses on harnessing the SegFormer segmentation model for precise semantic segmentation of strawberry diseases, aiming to improve disease detection accuracy under natural acquisition conditions.Methods: Three distinct Mix Transformer encoders - MiT-B0, MiT-B3, and MiT-B5 - were thoroughly analyzed to enhance disease detection, targeting diseases such as Angular leaf spot, Anthracnose rot, Blossom blight, Gray mold, Leaf spot, Powdery mildew on fruit, and Powdery mildew on leaves. The dataset consisted of 2,450 raw images, expanded to 4,574 augmented images. The Segment Anything Model integrated into the Roboflow annotation tool facilitated efficient annotation and dataset preparation.Results: The results reveal that MiT-B0 demonstrates balanced but slightly overfitting behavior, MiT-B3 adapts rapidly with consistent training and validation performance, and MiT-B5 offers efficient learning with occasional fluctuations, providing robust performance. MiT-B3 and MiT-B5 consistently outperformed MiT-B0 across disease types, with MiT-B5 achieving the most precise segmentation in general.Discussion: The findings provide key insights for researchers to select the most suitable encoder for disease detection applications, propelling the field forward for further investigation. The success in strawberry disease analysis suggests potential for extending this approach to other crops and diseases, paving the way for future research and interdisciplinary collaboration.

The increasing broiler demand due to overpopulation and meat imports presents challenges in poultry farming, including management, disease control, and chicken observation in varying light conditions. To address these issues, the development of AI-based management processes is crucial, especially considering the need for detecting pathological phenomena in intensive rearing. In this study, a dataset consisting of visual and thermal images was created to capture pathological phenomena in broilers. The dataset contains 10,000 images with 50,000 annotations labeled as lethargic chickens, slipped tendons, diseased eyes, stressed (beaks open), pendulous crop, and healthy broiler. Three versions of the YOLO-based algorithm (v8, v7, and v5) were assessed, utilizing augmented thermal and visual image datasets with various augmentation methods. The aim was to develop thermal- and visual-based models for detecting broilers in complex environments, and secondarily, to classify pathological phenomena under challenging lighting conditions. After training on acknowledged pathological phenomena, the thermal YOLOv8-based model demonstrated exceptional performance, achieving the highest accuracy in object detection (mAP50 of 0.988) and classification (F1 score of 0.972). This outstanding performance makes it a reliable tool for both broiler detection and pathological phenomena classification, attributed to the use of comprehensive datasets during training and development, enabling accurate and efficient detection even in complex environmental conditions. By employing both visual- and thermal-based models for monitoring, farmers can obtain results from both thermal and visual viewpoints, ultimately enhancing the overall reliability of the monitoring process.

Fresh tomato fruits (TFs) contain a high moisture content of 90–94%, which makes storage and transportation over long distances difficult. Lately, numerous investigators have employed diverse solar dryers (SDs) in conjunction with stationary solar collectors (SCs) to dry tomatoes; however, the effectiveness of this technique is limited due to the sun’s constant motion throughout the day. Consequently, the current study set out to create an SD that is outfitted with an autonomous sun tracking system and an internet of things (IoT)-based photovoltaic system connected to an SC to continually track the sun and increase the quantity of energy absorbed. Furthermore, we investigated some operating parameters that impact the SD’s performance, taking into account three tomato slice thicknesses (STs) (4.0, 6.0, and 8.0 mm) and three air velocities (1.0, 1.5, and 2.0 m/s). The obtained data demonstrated a notable rise in the efficiency of the SD integrated with the automatic SC tracker throughout the course of the day when compared to the fixed SC, where the latter’s efficiency improved by 21.6%, indicating a strong degree of agreement. The results demonstrated a notable 20–25% reduction in drying time and a 4.9 °C increase in air temperature within the SC integrated with an automatic solar collector tracker (ASCT) at 2:00 p.m., as compared to the SC integrated with a fixed SC. The results of this study also demonstrated that there were no appreciable variations in the air speeds used to dry the tomatoes; however, the thickness of the tomato slices (TSs) had a significant impact; using 4 mm thick tomato slices resulted in a 50% reduction in drying time. Furthermore, the highest efficiency of the PV system was discovered to be 17.45%. Although the two solar dryers have very similar payback times, there are more dried tomatoes available in the markets.

Optimizing feeding rates in recirculating aquaculture systems (RAS) is crucial for ensuring fish growth, health, and system efficiency. This research introduces a novel approach to RAS feeding control using a Deep Deterministic Policy Gradient (DDPG) reinforcement learning algorithm. The developed system integrates feeding rate optimization with comprehensive water quality management to improve overall performance and stability. The DDPG controller demonstrated superior tracking accuracy, reduced feed consumption, and improved operational stability compared to traditional control methods such as Model Predictive Control (MPC), PID, and Bang-Bang control. The learned policy-maintained feeding rates within optimal ranges while adapting to dynamic system requirements and environmental conditions. The integration of water quality monitoring and control further enhanced system stability, ensuring critical parameters remained within target ranges. Comparative analysis revealed the DDPG controller’s advantages in terms of faster recovery times after environmental perturbations, improved long-term stability, and significant economic benefits through reduced operational costs and increased efficiency. The robustness and adaptability of the system were validated through comprehensive testing under various fault conditions, growth phases, and system scales. The successful development and evaluation of the DDPG-based RAS feeding control system represents a significant advancement in aquaculture management. The demonstrated improvements in efficiency, stability, and economic performance establish the potential for this approach to revolutionize feeding practices in commercial RAS operations. Further research and development efforts can build upon these findings to advance the state-of-the-art in intelligent aquaculture management systems and promote the sustainability and profitability of the industry.

Recirculating Aquaculture Systems (RAS) have emerged as a sustainable approach to intensive fish production. In our previous work, we demonstrated the efficacy of a Deep Deterministic Policy Gradient (DDPG) reinforcement learning approach for optimizing feeding rates and maintaining water quality in RAS. However, the “black box” nature of deep reinforcement learning algorithms often impedes their adoption by fish farmers who lack specialized knowledge in artificial intelligence. This study addresses this gap by developing and evaluating methodologies to enhance the interpretability and explainability of DDPG-based control systems specifically for aquaculture practitioners. We employ decision tree approximation techniques to transform complex neural network policies into transparent rule-based systems while preserving performance. Our approach achieves 92.7% policy fidelity with only a 5.8% performance drop compared to the original DDPG controller. The resulting decision rules were validated with fish farmers, with 87% reporting improved understanding of system recommendations and 78% expressing increased trust in the technology. Feature importance analysis revealed dissolved oxygen, ammonia levels, and biomass as the primary decision factors. This work demonstrates how advanced artificial intelligence techniques can be made accessible to end-users through interpretable approximations, potentially accelerating the adoption of intelligent control systems in aquaculture and contributing to more sustainable food production practices.

Recirculating aquaculture systems (RAS) represent a promising solution for sustainable fish production, but their commercial viability is hampered by a critical barrier: adapting intelligent control systems to new fish species requires extensive, species-specific data collection and lengthy retraining periods (45–60 days). This challenge creates significant economic and operational hurdles for multi-species facilities, limiting their flexibility to adapt to market demands. This study addresses this fundamental limitation by introducing a novel framework that integrates transfer learning and federated intelligence to enable rapid, cost-effective, cross-species adaptation of deep reinforcement learning controllers. Building on our previous work with deep deterministic policy gradient (DDPG), we developed a modular neural architecture with species-agnostic and species-specific components. The system was validated across five distinct RAS configurations using three commercially important species: tilapia, rainbow trout, and European sea bass. The framework achieved 87.3% of optimal performance for a new species with just 14 days of adaptation data, a dramatic improvement over traditional approaches. Furthermore, the federated learning implementation enabled continuous, privacy-preserving model improvement across multiple facilities, demonstrating a 23.5% collective performance improvement over individually trained systems. Economic analysis confirmed the framework’s commercial viability, with adaptation costs 76% lower than developing new species-specific systems and a projected return on investment of 4–14 months. This research advances adaptive intelligent systems for aquaculture, offering a scalable and economically viable approach to precision RAS management. By significantly reducing implementation barriers, this work paves the way for wider commercial adoption, supporting the sustainable intensification required to meet global protein demands.