A Multi-Scale Target Detection Method Using an Improved Faster Region Convolutional Neural Network Based on Enhanced Backbone and Optimized Mechanisms.

Currently, existing deep learning methods exhibit many limitations in multi-target detection, such as low accuracy and high rates of false detection and missed detections. This paper proposes an improved Faster R-CNN algorithm, aiming to enhance the algorithm's capability in detecting multi-scale targets. This algorithm has three improvements based on Faster R-CNN. Firstly, the new algorithm uses the ResNet101 network for feature extraction of the detection image, which achieves stronger feature extraction capabilities. Secondly, the new algorithm integrates Online Hard Example Mining (OHEM), Soft non-maximum suppression (Soft-NMS), and Distance Intersection Over Union (DIOU) modules, which improves the positive and negative sample imbalance and the problem of small targets being easily missed during model training. Finally, the Region Proposal Network (RPN) is simplified to achieve a faster detection speed and a lower miss rate. The multi-scale training (MST) strategy is also used to train the improved Faster R-CNN to achieve a balance between detection accuracy and efficiency. Compared to the other detection models, the improved Faster R-CNN demonstrates significant advantages in terms of mAP@0.5, F1-score, and Log average miss rate (LAMR). The model proposed in this paper provides valuable insights and inspiration for many fields, such as smart agriculture, medical diagnosis, and face recognition.

SenseLite: A YOLO-Based Lightweight Model for Small Object Detection in Aerial Imagery.

In the field of aerial remote sensing, detecting small objects in aerial images is challenging. Their subtle presence against broad backgrounds, combined with environmental complexities and low image resolution, complicates identification. While their detection is crucial for urban planning, traffic monitoring, and military reconnaissance, many deep learning approaches demand significant computational resources, hindering real-time applications. To elevate the accuracy of small object detection in aerial imagery and cater to real-time requirements, we introduce SenseLite, a lightweight and efficient model tailored for aerial image object detection. First, we innovatively structured the YOLOv5 model for a more streamlined structure. In the backbone, we replaced the original structure with cutting-edge lightweight neural operator Involution, enhancing contextual semantics and weight distribution. For the neck, we incorporated GSConv and slim-Neck, striking a balance between reduced computational complexity and performance, which is ideal for rapid predictions. Additionally, to enhance detection accuracy, we integrated a squeeze-and-excitation (SE) mechanism to amplify channel communication and improve detection accuracy. Finally, the Soft-NMS strategy was employed to manage overlapping targets, ensuring precise concurrent detections. Performance-wise, SenseLite reduces parameters by 30.5%, from 7.05 M to 4.9 M, as well as computational demands, with GFLOPs decreasing from 15.9 to 11.2. It surpasses the original YOLOv5, showing a 5.5% mAP0.5 improvement, 0.9% higher precision, and 1.4% better recall on the DOTA dataset. Compared to other leading methods, SenseLite stands out in terms of performance.

Research on Apple Recognition Algorithm in Complex Orchard Environment Based on Deep Learning.

In the complex environment of orchards, in view of low fruit recognition accuracy, poor real-time and robustness of traditional recognition algorithms, this paper propose an improved fruit recognition algorithm based on deep learning. Firstly, the residual module was assembled with the cross stage parity network (CSP Net) to optimize recognition performance and reduce the computing burden of the network. Secondly, the spatial pyramid pool (SPP) module is integrated into the recognition network of the YOLOv5 to blend the local and global features of the fruit, thus improving the recall rate of the minimum fruit target. Meanwhile, the NMS algorithm was replaced by the Soft NMS algorithm to enhance the ability of identifying overlapped fruits. Finally, a joint loss function was constructed based on focal and CIoU loss to optimize the algorithm, and the recognition accuracy was significantly improved. The test results show that the MAP value of the improved model after dataset training reaches 96.3% in the test set, which is 3.8% higher than the original model. F1 value reaches 91.8%, which is 3.8% higher than the original model. The average detection speed under GPU reaches 27.8 frames/s, which is 5.6 frames/s higher than the original model. Compared with current advanced detection methods such as Faster RCNN and RetinaNet, among others, the test results show that this method has excellent detection accuracy, good robustness and real-time performance, and has important reference value for solving the problem of accurate recognition of fruit in complex environment.

Real-Time Vehicle Detection from UAV Aerial Images Based on Improved YOLOv5.

Aerial vehicle detection has significant applications in aerial surveillance and traffic control. The pictures captured by the UAV are characterized by many tiny objects and vehicles obscuring each other, significantly increasing the detection challenge. In the research of detecting vehicles in aerial images, there is a widespread problem of missed and false detections. Therefore, we customize a model based on YOLOv5 to be more suitable for detecting vehicles in aerial images. Firstly, we add one additional prediction head to detect smaller-scale objects. Furthermore, to keep the original features involved in the training process of the model, we introduce a Bidirectional Feature Pyramid Network (BiFPN) to fuse the feature information from various scales. Lastly, Soft-NMS (soft non-maximum suppression) is employed as a prediction frame filtering method, alleviating the missed detection due to the close alignment of vehicles. The experimental findings on the self-made dataset in this research indicate that compared with YOLOv5s, the mAP@0.5 and mAP@0.5:0.95 of YOLOv5-VTO increase by 3.7% and 4.7%, respectively, and the two indexes of accuracy and recall are also improved.

YOLOv5s-Fog: An Improved Model Based on YOLOv5s for Object Detection in Foggy Weather Scenarios.

In foggy weather scenarios, the scattering and absorption of light by water droplets and particulate matter cause object features in images to become blurred or lost, presenting a significant challenge for target detection in autonomous driving vehicles. To address this issue, this study proposes a foggy weather detection method based on the YOLOv5s framework, named YOLOv5s-Fog. The model enhances the feature extraction and expression capabilities of YOLOv5s by introducing a novel target detection layer called SwinFocus. Additionally, the decoupled head is incorporated into the model, and the conventional non-maximum suppression method is replaced with Soft-NMS. The experimental results demonstrate that these improvements effectively enhance the detection performance for blurry objects and small targets in foggy weather conditions. Compared to the baseline model, YOLOv5s, YOLOv5s-Fog achieves a 5.4% increase in mAP on the RTTS dataset, reaching 73.4%. This method provides technical support for rapid and accurate target detection in adverse weather conditions, such as foggy weather, for autonomous driving vehicles.

Multi-scenario pear tree inflorescence detection based on improved YOLOv7 object detection algorithm.

Efficient and precise thinning during the orchard blossom period is a crucial factor in enhancing both fruit yield and quality. The accurate recognition of inflorescence is the cornerstone of intelligent blossom equipment. To advance the process of intelligent blossom thinning, this paper addresses the issue of suboptimal performance of current inflorescence recognition algorithms in detecting dense inflorescence at a long distance. It introduces an inflorescence recognition algorithm, YOLOv7-E, based on the YOLOv7 neural network model. YOLOv7 incorporates an efficient multi-scale attention mechanism (EMA) to enable cross-channel feature interaction through parallel processing strategies, thereby maximizing the retention of pixel-level features and positional information on the feature maps. Additionally, the SPPCSPC module is optimized to preserve target area features as much as possible under different receptive fields, and the Soft-NMS algorithm is employed to reduce the likelihood of missing detections in overlapping regions. The model is trained on a diverse dataset collected from real-world field settings. Upon validation, the improved YOLOv7-E object detection algorithm achieves an average precision and recall of 91.4% and 89.8%, respectively, in inflorescence detection under various time periods, distances, and weather conditions. The detection time for a single image is 80.9 ms, and the model size is 37.6 Mb. In comparison to the original YOLOv7 algorithm, it boasts a 4.9% increase in detection accuracy and a 5.3% improvement in recall rate, with a mere 1.8% increase in model parameters. The YOLOv7-E object detection algorithm presented in this study enables precise inflorescence detection and localization across an entire tree at varying distances, offering robust technical support for differentiated and precise blossom thinning operations by thinning machinery in the future.

Automatic Detection and Segmentation for Group-Housed Pigs Based on PigMS R-CNN.

Instance segmentation is an accurate and reliable method to segment adhesive pigs' images, and is critical for providing health and welfare information on individual pigs, such as body condition score, live weight, and activity behaviors in group-housed pig environments. In this paper, a PigMS R-CNN framework based on mask scoring R-CNN (MS R-CNN) is explored to segment adhesive pig areas in group-pig images, to separate the identification and location of group-housed pigs. The PigMS R-CNN consists of three processes. First, a residual network of 101-layers, combined with the feature pyramid network (FPN), is used as a feature extraction network to obtain feature maps for input images. Then, according to these feature maps, the region candidate network generates the regions of interest (RoIs). Finally, for each RoI, we can obtain the location, classification, and segmentation results of detected pigs through the regression and category, and mask three branches from the PigMS R-CNN head network. To avoid target pigs being missed and error detections in overlapping or stuck areas of group-housed pigs, the PigMS R-CNN framework uses soft non-maximum suppression (soft-NMS) by replacing the traditional NMS to conduct post-processing selected operation of pigs. The MS R-CNN framework with traditional NMS obtains results with an F1 of 0.9228. By setting the soft-NMS threshold to 0.7 on PigMS R-CNN, detection of the target pigs achieves an F1 of 0.9374. The work explores a new instance segmentation method for adhesive group-housed pig images, which provides valuable exploration for vision-based, real-time automatic pig monitoring and welfare evaluation.