Accurate recognition of fruits in the orchard is an important step for robot picking in the natural environment, since many CNN models have a low recognition rate when dealing with irregularly shaped and very dense fruits, such as a grape bunch. It is a new trend to use a transformer structure and apply it to a computer vision domain for image processing. This paper provides Swin Transformer and DETR models to achieve grape bunch detection. Additionally, they are compared with traditional CNN models, such as Faster-RCNN, SSD, and YOLO. In addition, the optimal number of stages for a Swin Transformer through experiments is selected. Furthermore, the latest YOLOX model is also used to make a comparison with the Swin Transformer, and the experimental results show that YOLOX has higher accuracy and better detection effect. The above models are trained under red grape datasets collected under natural light. In addition, the dataset is expanded through image data augmentation to achieve a better training effect. After 200 epochs of training, SwinGD obtained an exciting mAP value of 94% when IoU = 0.5. In case of overexposure, overdarkness, and occlusion, SwinGD can recognize more accurately and robustly compared with other models. At the same time, SwinGD still has a better effect when dealing with dense grape bunches. Furthermore, 100 pictures of grapes containing 655 grape bunches are downloaded from Baidu pictures to detect the effect. The Swin Transformer has an accuracy of 91.5%. In order to verify the universality of SwinGD, we conducted a test under green grape images. The experimental results show that SwinGD has a good effect in practical application. The success of SwinGD provides a new solution for precision harvesting in agriculture.

Counting grape berries and measuring their size can provide accurate data for robot picking behavior decision-making, yield estimation, and quality evaluation. When grapes are picked, there is a strong uncertainty in the external environment and the shape of the grapes. Counting grape berries and measuring berry size are challenging tasks. Computer vision has made a huge breakthrough in this field. Although the detection method of grape berries based on 3D point cloud information relies on scanning equipment to estimate the number and yield of grape berries, the detection method is difficult to generalize. Grape berry detection based on 2D images is an effective method to solve this problem. However, it is difficult for traditional algorithms to accurately measure the berry size and other parameters, and there is still the problem of the low robustness of berry counting. In response to the above problems, we propose a grape berry detection method based on edge image processing and geometric morphology. The edge contour search and the corner detection algorithm are introduced to detect the concave point position of the berry edge contour extracted by the Canny algorithm to obtain the best contour segment. To correctly obtain the edge contour information of each berry and reduce the error grouping of contour segments, this paper proposes an algorithm for combining contour segments based on clustering search strategy and rotation direction determination, which realizes the correct reorganization of the segmented contour segments, to achieve an accurate calculation of the number of berries and an accurate measurement of their size. The experimental results prove that our proposed method has an average accuracy of 87.76% for the detection of the concave points of the edge contours of different types of grapes, which can achieve a good edge contour segmentation. The average accuracy of the detection of the number of grapes berries in this paper is 91.42%, which is 4.75% higher than that of the Hough transform. The average error between the measured berry size and the actual berry size is 2.30 mm, and the maximum error is 5.62 mm, which is within a reasonable range. The results prove that the method proposed in this paper is robust enough to detect different types of grape berries.

Energy consumption has become a major concern to the widespread deployment of cloud data centers. Many techniques have been devised to help reduce energy consumption for cloud data centers that consist of a large number of identical servers, including dynamic allocation of active servers, consolidating diverse applications, and adjusting the CPU frequency of an active server. However, these techniques normally have a high migration and low resource utilization. CPU and memory is the dominant factors of the performance and energy consumption and whose allocation determines the energy efficiency of cloud system. Leveraging these techniques, we focus on the problem of VM placement, propose a heuristic greedy algorithm to implement VM deployment and live migration to maximize total resource utilization and minimize energy consumption, which is based on energy-aware and quadratic exponential smoothing method to predict the workloads. Our heuristic algorithm makes CPU-intensive services and memory-intensive services mapped to the same physical server more complementary. The experiment results show that there is significant improvement in the aspect of energy saving, workload balancing and scalability, compared with single-objective approaches based on CPU utilization.

Speech translation (ST) is a bimodal conversion task from source speech to the target text. Generally, deep learning-based ST systems require sufficient training data to obtain a competitive result, even with a state-of-the-art model. However, the training data is usually unable to meet the completeness condition due to the small sample problems. Most low-resource ST tasks improve data integrity with a single model, but this optimization has a single dimension and limited effectiveness. In contrast, multimodality is introduced to leverage different dimensions of data features for multiperspective modeling. This approach mutually addresses the gaps in the different modalities to enhance the representation of the data and improve the utilization of the training samples. Therefore, it is a new challenge to leverage the enormous multimodal out-of-domain information to improve the low-resource tasks. This paper describes how to use multimodal out-of-domain information to improve low-resource models. First, we propose a low-resource ST framework to reconstruct large-scale label-free audio by combining self-supervised learning. At the same time, we introduce a machine translation (MT) pretraining model to complement text embedding and fine-tune decoding. In addition, we analyze the similarity at the decoder side. We reduce multimodal invalid pseudolabels by performing random depth pruning in the similarity layer to minimize error propagation and use additional CTC loss in the nonsimilarity layer to optimize the ensemble loss. Finally, we study the weighting ratio of the fusion technique in the multimodal decoder. Our experiment results show that the proposed method is promising for low-resource ST, with improvements of up to +3.6 BLEU points compared to baseline low-resource ST models.

The non‐independent and identically distributed (non‐IID) nature of data across different sites and the dynamic participation of client nodes represent two critical challenges faced by distributed industrial visual defect detection systems. While asynchronous federated learning (AFL) reduces training inefficiencies, its asynchronous update process makes the global model more vulnerable to data heterogeneity. This research proposes a contrastive clustering–based asynchronous federated learning (CC‐AFL) framework. The method integrates contrastive learning into the client clustering process, working simultaneously across instance‐level and cluster‐level spaces to identify hidden similarity patterns among participating clients. The server maintains update queues for each cluster and then applies staleness‐weighted aggregation based on the combined timestamp. Meanwhile, the framework uses client scheduling based on cluster queue lengths, giving priority to clusters with fewer updates to balance update frequencies across different clusters. Without requiring raw data sharing, the framework achieves improvements in both convergence speed and stability of AFL under changing scenarios. On the furniture board dataset, CC‐AFL achieves better performance with an average area under the receiver operating characteristic curve (I‐AUROC) of 74.38%, showing a 2.22% improvement over AFL‐DCS under different non‐IID settings. Experimental results show that CC‐AFL maintains stable convergence under different conditions, confirming the effectiveness and practical value of contrastive clustering and scheduling in AFL.