Advancements in technology hold significant prom-ise for the future of smart agriculture. Drones, serving as innovative tools for data acquisition, play a pivotal role in this context. Traditional pollination drones, which rely on extensive spraying methods, suffer from poor efficiency. To address this challenge, our focus is on enhancing pollination through object detection technology, leading to the introduction of the YOLO-FT detection algorithm based on YOLOv7. To validate our algorithm, we have curated a dataset of fruit tree flowers. Our approach involves several key components. Firstly, we propose a lightweight adaptive backbone network, which effectively extracts feature information through partial convolution and seamlessly integrates feature data from diverse channels. Secondly, we intro-duce a non-parametric attention module into neck network, bolstering the fusion of critical feature information. Finally, we leverage a rapid convergence function based on centroid distance to enhance bounding box regression performance. Experimental results demonstrate the superiority of the YOLO-FT algorithm, achieving mAP50 of 92.8%, mAP50-95 of 48.8%, and F1 score of 0.90, marking improvements of 3.6%, 1.6%, and 0.17, respectively, over the baseline. Furthermore, YOLO-FT exhibits a signifi-cant reduction in model complexity, with parameters decreased by 13.4% and FLOPs by 17.8%. This research serves as a valuable theoretical reference for advancements in smart agriculture.

In intelligent transportation systems, accurate vehicle target recognition within road scenarios is crucial for achieving intelligent traffic management. Addressing the challenges posed by complex environments and severe vehicle occlusion in such scenarios, this paper proposes a novel vehicle-detection method, YOLO-BOS. First, to bolster the feature-extraction capabilities of the backbone network, we propose a novel Bi-level Routing Spatial Attention (BRSA) mechanism, which selectively filters features based on task requirements and adjusts the importance of spatial locations to more accurately enhance relevant features. Second, we incorporate Omni-directional Dynamic Convolution (ODConv) into the head network, which is capable of simultaneously learning complementary attention across the four dimensions of the kernel space, therefore facilitating the capture of multifaceted features from the input data. Lastly, we introduce Shape-IOU, a new loss function that significantly enhances the accuracy and robustness of detection results for vehicles of varying sizes. Experimental evaluations conducted on the UA-DETRAC dataset demonstrate that our model achieves improvements of 4.7 and 4.4 percentage points in mAP@0.5 and mAP@0.5:0.95, respectively, compared to the baseline model. Furthermore, comparative experiments on the SODA10M dataset corroborate the superiority of our method in terms of precision and accuracy.

To achieve accurate detection of tomato fruit maturity and enable automated harvesting in natural environments, this paper presents a more lightweight and efficient maturity detection algorithm, YOLO-DGS, addressing the challenges of subtle maturity differences between regular and cherry tomatoes, as well as fruit occlusion. First, to enhance feature extraction at various levels of abstraction in the input data, this paper proposes a novel segment-wise convolution module, C2f-GB. This module performs convolution in stages on the feature map, generating more feature maps with fewer parameters and computational resources, thereby improving the model’s feature extraction capability while reducing parameter count and computational cost. Next, based on the YOLO v10 algorithm, this paper removes redundant detection layers to enhance the model’s ability to capture specific features and further reduce the number of parameters. This paper then integrates a bidirectional feature pyramid network (BiFPN) into the neck network to improve feature capture across different scales, enhancing the model’s ability to handle objects of varying sizes and complexities. Finally, we introduce a novel channel attention mechanism that allows the network to dynamically adjust its focus on channels, efficiently utilizing available information. Experimental results demonstrate that the improved YOLO-DGS model achieves a 2.6% increase in F1 score, 2.1% in recall, 2% in mAP50, and 1% in mAP50-95. Additionally, inference speed is improved by 12.5%, and the number of parameters is reduced by 26.3%. Compared to current mainstream lightweight object detection models, YOLO-DGS outperforms them, offering an efficient solution for the tomato harvesting robot vision system in natural environments.

Automatically finding correspondences between object features in images is of main interest for several applications, as object detection and tracking, identification, registration, and many derived tasks. In this paper, we address feature correspondence within the general framework of graph matching optimization and with the principal aim to contribute. We proposed two optimized algorithms: first-order and second-order for graph matching. On the one hand, a first-order normalized cross-correlation (NCC) based graph matching algorithm using entropy and response through Marr wavelets within the scale-interaction method is proposed. First, we proposed a new automatic feature detection processing by using Marr wavelets within the scale-interaction method. Second, feature extraction is executed under the mesh division strategy and entropy algorithm, accompanied by the assessment of the distribution criterion. Image matching is achieved by the nearest neighbor search with normalized cross-correlation similarity measurement to perform coarse matching on feature points set. As to the matching points filtering part, the Random Sample Consensus Algorithm (RANSAC) removes outliers correspondences. One the other hand, a second-order NCC based graph matching algorithm is presented. This algorithm is an integer quadratic programming (IQP) graph matching problem, which is implemented in Matlab. It allows developing and comparing many algorithms based on a common evaluation platform, sharing input data, and a customizable affinity matrix and matching list of candidate solution pairs as input data. Experimental results demonstrate the improvements of these algorithms concerning matching recall and accuracy compared with other algorithms.