Traditional segmentation methods are slow and rely on manual annotations, which are labor-intensive. To address these limitations, we propose YOLO-SAM AgriScan, a unified framework that combines the fast object detection capabilities of YOLOv11 with the zero-shot segmentation power of the Segment Anything Model 2 (SAM2). Our approach adopts a hybrid paradigm for on-plant ripe strawberry segmentation, wherein YOLOv11 is fine-tuned using a few-shot learning strategy with minimal annotated samples, and SAM2 performs mask generation without additional supervision. This architecture eliminates the bottleneck of pixel-wise manual annotation and enables the scalable and efficient segmentation of strawberries in both controlled and natural farm environments. Experimental evaluations on two datasets, a custom-collected dataset and a publicly available benchmark, demonstrate strong detection and segmentation performance in both full-data and data-constrained scenarios. The proposed framework achieved a mean Dice score of 0.95 and an IoU of 0.93 on our collected dataset and maintained competitive performance on public data (Dice: 0.95, IoU: 0.92), demonstrating its robustness, generalizability, and practical relevance in real-world agricultural settings. Our results highlight the potential of combining few-shot detection and zero-shot segmentation to accelerate the development of annotation-light, intelligent phenotyping systems.

Recent advancements in 3D reconstruction technologies have significantly transformed plant phenotyping, enabling precise, scalable, and automated trait extraction. Traditional manual phenotyping methods are increasingly being replaced by image-based approaches, such as photogrammetry, LiDAR, RGB-D sensing, and deep learning (DL)-based techniques. These tools allow for non-destructive, high-throughput measurements of plant morphology, structure, and physiological traits. This review synthesizes the state of the art in 3D reconstruction methods, including conventional geometric algorithms and emerging DL methods, and evaluates their application across diverse plant species. In addition, we discuss the sensing modalities, evaluation metrics, and crop-specific deployments. Although promising, current technologies still face challenges in terms of computational efficiency, scalability to outdoor environments, and generalizability across crop types. This review concludes by identifying research gaps and future directions for making real-time, field-deployable 3D phenotyping systems.

Highlights An end-effector prototype was developed for harvesting greenhouse hydroponic lettuce. The end-effector's cutting mechanism was designed for 1.72 Nm torque, accounting for 25% loss. The harvesting test showed that the end-effector can cut and hold simultaneously with 95.15% and 94.45% success rates. The average remnant leaves per plant was 0.55 after cutting and fallen leaves per plant was 0.075 at maneuvering. ABSTRACT. Lettuce is one of the most demanding annual leafy green vegetables. Greenhouse lettuce, meanwhile, is at a much earlier stage of its evolution, with a cost of production still significantly higher than that of field-grown lettuce. Labor accounts for nearly one-third of the total production cost. Decreasing labor availability and increasing labor costs are the two most critical challenges in greenhouse lettuce production. Thus, automation to reduce labor requirements is required for the industry's sustainability. This study aims to develop a prototype of a robotic end-effector for harvesting greenhouse hydroponic lettuce. Lettuce cutting force was obtained using a thin force sensor, which was used to measure the cutting torque to provide guidelines for the end-effector design. The test result suggested that 1.72 Nm torque was required to cut the lettuce stem. An end-effector prototype with a pair of shear blades and a two-finger gripper mechanism operated using a single linear actuator was designed. Cutting and holding mechanisms work together using the first-class lever technique, where applied and acting forces are connected at the pivot point. The end-effector prototype was integrated into a linear arm attached to a two-directional linear manipulator system operated using an embedded controller. Validation experiments of the end-effector for harvesting lettuce growing in a Nutrient Film Technique (NFT) hydroponic system were conducted. The results indicated a 95.15% success rate in cutting and a 94.45% success rate in securely holding the lettuce during harvest. The developed robotic end-effector serves as a pivotal component in lettuce harvesting automation. Keywords: Controlled environment agriculture, Greenhouse lettuce, Leafy green harvesting, Precision agriculture, Robotic end-effector, Stem cutting force.

This study presents a deep-learning-based vision system for segmenting lettuce plants and localizing cutting points, enabling robotic harvesting in greenhouse hydroponics. A total of 784 RGB-D (Color-Depth) images were captured for lettuce segmentation. An additional 80 images were collected and annotated to establish the ground truth for cutting point localization. Image preprocessing generated two datasets: one with original RGB images and another with background removed via depth thresholding. Two instance segmentation models, YOLO11-seg and Mask R-CNN (Region-based Convolutional Neural Network) in Detectron2 framework, were trained to segment individual lettuce plants and the nutrient film technique hydroponic channel. An algorithm was developed to estimate the stem and determine the occluded cutting point from the segmentation masks. The segmentation results indicated that YOLO11-seg outperformed Mask R-CNN, achieving 99.09% precision, 99.76% recall, and 97.53% mean average precision, at an intersection over union of 75%. The cutting-point localization algorithm was evaluated using ground-truth images, achieved an accuracy of 85% and demonstrating strong agreement with the ground-truth data (R$^{2}$ = 0.99 and root mean squared error = 11.95 $\pm$ 0.05 mm). These results validate the effectiveness of our proposed vision system in enabling autonomous harvesting of hydroponic lettuce.

Biometrics is a technology used to identify, analyze, and measure an individual’s physical and behavioral characteristics. It is used for authenticating and authorizing a person. Among all other biometric authentication, fingerprint recognition is the most known and used solution to authenticate people on biometric systems. Usually, fingerprint recognition approaches are minutiae-based and correlation-based. However, the minutiae-based approach is popular and extensively used method for fingerprint authentication, it shows poor performance for low quality images and insecure over data-passing channel. In proposed methodology, feature-based approach for fingerprint recognition is developed by enhancing the low quality image using FFT and Gaussian filter and by extracting the feature of the image using Gabor filter. Similarity measurement is done by calculating the cosine-similarity value of correlation factors. The cosine-similarity value of correlation factors of input image and template image are computed and compared. If it is over a certain threshold the result of the matching process is positive otherwise negative.

Accurate measurement of vapor pressure deficit (VPD) at the leaf surface is essential for assessing plant water status and stress. However, current greenhouse climate control strategies that rely on ambient sensors for air VPD measurements do not capture the localized microclimate at the leaf surface, limiting their usefulness for plant-centered stress detection. This study presents a low-cost, plant-integrated sensing platform for real-time, leaf-level VPD monitoring. A flexible temperature and humidity sensor suite was mounted directly on lettuce leaves grown under controlled hydroponic greenhouse conditions, enabling continuous measurement of the stomatal boundary layer microenvironment. Leaf-level temperature and relative humidity (RH) data were used to compute real-time VPD and assess plant responses under well-watered and water-stressed conditions. Sensor measurements showed that water-stressed plants consistently exhibited higher leaf-surface temperatures (~0.71 °C above controls) and lower leaf-surface RH (~8% below controls), reflecting reduced transpiration-driven cooling. Calculated VPD values ranged from 0.7-1.69 kPa under water stress compared to 0.5-1.35 kPa in control plants, with a consistent treatment difference of approximately 0.2-0.34 kPa. These trends in sensor data were validated by independent measurements using a CIRAS-4 photosynthesis system, which confirmed higher VPD and reduced stomatal conductance under stress. In contrast, ambient sensors failed to resolve these differences across treatments. These findings demonstrate that leaf-integrated sensors can bridge the gap between ambient measurements and the true leaf microclimate. The proposed platform offers a scalable and non-invasive approach for real-time VPD monitoring at the leaf-scale, supporting improved drought stress detection, irrigation scheduling, and precision climate control decisions in a greenhouse.