Traditional Object Detection Methods Research Articles

Identifying rice field weeds from unmanned aerial vehicle remote sensing imagery using deep learning

BackgroundRice field weed object detection can provide key information on weed species and locations for precise spraying, which is of great significance in actual agricultural production. However, facing the complex and changing real farm environments, traditional object detection methods still have difficulties in identifying small-sized, occluded and densely distributed weed instances. To address these problems, this paper proposes a multi-scale feature enhanced DETR network, named RMS-DETR. By adding multi-scale feature extraction branches on top of DETR, this model fully utilizes the information from different semantic feature layers to improve recognition capability for rice field weeds in real-world scenarios.MethodsIntroducing multi-scale feature layers on the basis of the DETR model, we conduct a differentiated design for different semantic feature layers. The high-level semantic feature layer adopts Transformer structure to extract contextual information between barnyard grass and rice plants. The low-level semantic feature layer uses CNN structure to extract local detail features of barnyard grass. Introducing multi-scale feature layers inevitably leads to increased model computation, thus lowering model inference speed. Therefore, we employ a new type of Pconv (Partial convolution) to replace traditional standard convolutions in the model.ResultsCompared to the original DETR model, our proposed RMS-DETR model achieved an average recognition accuracy improvement of 3.6% and 4.4% on our constructed rice field weeds dataset and the DOTA public dataset, respectively. The average recognition accuracies reached 0.792 and 0.851, respectively. The RMS-DETR model size is 40.8 M with inference time of 0.0081 s. Compared with three classical DETR models (Deformable DETR, Anchor DETR and DAB-DETR), the RMS-DETR model respectively improved average precision by 2.1%, 4.9% and 2.4%.DiscussionThis model is capable of accurately identifying rice field weeds in complex real-world scenarios, thus providing key technical support for precision spraying and management of variable-rate spraying systems.

Depth-enhancement network for monocular 3D object detection

Abstract To accurately obtain 3D information, the correct use of depth data is crucial. Compared with radar-based methods, detecting objects in 3D space in a single image is extremely challenging due to the lack of depth cues. However, monocular 3D object detection provides a more economical solution. Traditional monocular 3D object detection methods often rely on geometric constraints, such as key points, object shape relationships and 3D to 2D optimization, to address the inherent lack of depth information. However, these methods still make it challenging to extract rich information directly from depth estimation for fusion. To fundamentally enhance the ability of monocular 3D object detection, we propose a monocular 3D object detection network based on depth information enhancement. The network learns object detection and depth estimation tasks simultaneously through a unified framework, integrates depth features as auxiliary information into the detection branch, and then constrains and enhances them to obtain better spatial representation. To this end, we introduce a new cross-modal fusion strategy, which realizes a more reasonable fusion of cross-modal information by exploring redundant, complementary information and their interactions between RGB features and depth features. Extensive experiments on the KITTI dataset show that our method can significantly improve the performance of monocular 3D object detection.

A dynamically class-wise weighting mechanism for unsupervised cross-domain object detection under universal scenarios

In the realm of object detection, traditional domain adaptive object detection (DAOD) methods assume that source and target data completely share one identical class space, which is often difficult to satisfy in many real-world applications. To address this limitation, this paper introduces universal domain adaptive object detection (UniDAOD), a learning paradigm that relaxes identical class space assumption to be a different but overlapped class space. Intuitively, the main challenge of UniDAOD is to reduce the negative transfer of private classes (i.e., classes only existed in one domain) and reinforce the positive transfer of the common classes (i.e., classes shared across domains). In this paper, we provide a rigorous theoretical analysis and induce a new generalization bound of the expected target error under the UniDAOD setting. On the basis of theoretical insight, we then propose weighted adaptation (W-adapt) to suppress the interference of private classes and reinforce the positive effects of common classes. In particular, we propose a pseudo category margin (PCM) to quantify class importance based on dynamic pseudotarget label prediction to recognize common classes. Furthermore, to alleviate the impact of inaccurate pseudotarget labels, we propose a temporary memory-based filter (TMF) to dynamically store and update the PCM during progressive training. On the basis of the learned TMF, we design a weighted classwise domain alignment loss to adapt two domains across common classes. Experiments on four universal scenarios (i.e., partial-set, open-partial-set, open-set, and closed-set) show that W-adapt outperforms several domain adaptation methods.

Enhancing Safety and Workover Efficiency Through Real-Time Scale-Adaptive Video Detection of Critical Objects at the Wellsite

Summary The wellsite serves as the fundamental unit in the development of oil and gas fields, functioning as a hub for the production activities, with workover operations being a critical means to ensure production continuity. In addition, it plays a crucial role in environmental protection, preventing oil and gas leakage and pollution. Various pieces of mechanical equipment deployed at the wellsite are essential for tasks such as oil and gas extraction and well repair operations, holding a pivotal position in oil- and gasfield development. Consequently, intelligent wellsite implementation necessitates a primary focus on monitoring mechanical equipment, with video emerging as a vital form of multisource information at the wellsite. While existing research on wellsite video monitoring predominantly addresses system and data transmission issues, it falls short in addressing the challenges of real-time assessment and early warning in intelligent wellsite operations. This study introduces a method for identifying critical targets at the wellsite based on a scale-adaptive network. The model employs a multiscale fusion network to extract different image features and semantic features at various scales, facilitating their fusion. The processing of wellsite video images occurs in multiple stages, outputting predicted box locations and category information, enabling the localization and recognition of critical objects at the wellsite. Unlike traditional deep convolutional object detection methods, this model incorporates a parameter-free attention mechanism, enhancing the accurate feature learning of small targets during the extraction process and addressing the issue of multiscale imbalance. The experimental results validate the robust performance of the method, surpassing the latest one-stage object detection models and mainstream loss function methods. Comparative experiments demonstrate a 9.22% improvement in mean average precision (mAP) compared with YOLOv8, establishing the proposed model as a top performer in loss function optimization experiments. Furthermore, we propose a video security detection model whose results, combined with the recognition model, are applicable for video detection in wellsite scenarios. The model exhibits strong integration capabilities for scene area division and behavior regulation monitoring. In addition, the model provides valuable insights for analyzing equipment operating status, aligning with the practical needs of oil fields.

Active Visual Perception Enhancement Method Based on Deep Reinforcement Learning

Traditional object detection methods using static cameras are constrained by their limited perspectives, hampering the effective detection of low-confidence targets. To address this challenge, this study introduces a deep reinforcement learning-based visual perception enhancement technique. This approach leverages pan–tilt–zoom (PTZ) cameras to achieve active vision, enabling them to autonomously make decisions and actions tailored to the current scene and object detection outcomes. This optimization enhances both the object detection process and information acquisition, significantly boosting the intelligent perception capabilities of PTZ cameras. Experimental findings demonstrate the robust generalization capabilities of this method across various object detection algorithms, resulting in an average confidence level improvement of 23.80%.

A structural description of pedestrian movement behavior in multiple surveillance videos

This study purpose of recording pedestrians' activities in all surveillance videos and then convert these records into structured descriptions to analyze a large number of surveillance videos. This paper proposes a four-stage processing method. The first stage uses convolutional neural network (CNN) to detect objects in the surveillance video and classify them. The second stage finds the correlation of pedestrians in continuous images and describes multiple objects' trajectories in the surveillance video. The third stage calculates the similarity of the pedestrian object sets of the surveillance video at different times as the basis for pedestrian re-identification. In the fourth stage, extract the walking path of the pedestrian in each surveillance video. Each pedestrian's actions in each surveillance video, including whether to stand, walking path, direction, clothing color and walking time. To construct a correlation table of pedestrian motion in multiple surveillance videos.This experiment uses two image datasets for experimentation and analysis to evaluate this method's effectiveness for pedestrian movement in multiple surveillance videos. The image datasets are the Multi-Camera Object Tracking (MCT) Challenge and PETS 2009 Benchmark Data. This experiment includes pedestrian detection, pedestrian identification, and multi-pedestrian tracking of pedestrians in single and multiple surveillance cameras. The results show that in the pedestrian detection experiment, the convolutional neural network can effectively solve the problems caused by static objects, light changes, similar colors, crowded pedestrians, and incomplete objects. In the pedestrian recognition and re-recognition experiments, the accuracy is reduced due to the influence of low-resolution surveillance videos. This detection method of the pedestrian is more accurate than traditional dynamic object detection methods. After the pedestrian tracking and re-identification are completed, the pedestrian relationships in multiple surveillance cameras are integrated, and a pedestrian behavior association table is built for surveillance videos.

FCFormer: fish density estimation and counting in recirculating aquaculture system

In intelligent feeding recirculating aquaculture system, accurately estimating fish population and density is pivotal for management practices and survival rate assessments. However, challenges arise due to mutual occlusion among fish, rapid movement, and complex breeding environments. Traditional object detection methods based on convolutional neural networks (CNN) often fall short in fully addressing the detection demands for fish schools, especially for distant and small targets. In this regard, we introduce a detection framework dubbed FCFormer (Fish Count Transformer). Specifically, the Twins-SVT backbone network is employed first to extract global features of fish schools. To further enhance feature extraction, especially in the fusion of features at different levels, a Bi-FPN aggregation network model with a CAM Count module is incorporated (BiCC). The CAM module aids in focusing more on critical region features, thus rendering feature fusion more cohesive and effective. Furthermore, to precisely predict density maps and elevate the accuracy of fish counting, we devised an adaptive feature fusion regression head: CRMHead. This approach not only optimizes the feature fusion process but also ensures superior counting precision. Experimental results shown that the proposed FCFormer network achieves an accuracy of 97.06%, with a mean absolute error (MAE) of 6.37 and a root mean square error (MSE) of 8.69. Compared to the Twins transformer, there's a 2.02% improvement, outperforming other transformer-based architectures like CCTrans and DM_Count. The presented FCFormer algorithm can be effectively applied to fish density detection in intelligent feeding recirculating aquaculture system, offering valuable input for the development of intelligent breeding management systems.

Improving the YOLOV7 Algorithm for Object Detection within Recorded Videos

Object detection algorithms play an important role in detecting people in surveillance videos. In recent years, with the rapid development of deep learning, the performance of object detection has improved by leaps and bounds, and the scheme of object detection by the YOLOV7 algorithm has also been born. Traditional object detection methods often fail to achieve a balance between speed and accuracy. To address these issues, in this research, an improved YOLOv7 algorithm performance is proposed to get the best speed-to-accuracy balance compared to state-of-the-art object detection within recorded videos using an effective compression method. This method calculates the difference between frames of video, and by using the zero difference approach by removing the duplicate frames from the recorded video and choosing only the meaningful frames based on many variables, including frame size, frame details, and the distance of the frames, influence the choice of a meaningful frame, and this will reduce the size of the video by eliminating the frames comparable to those chosen. Additionally, any other datasets or pre-trained weights have not been used; YOLOv7 has been exclusively trained on the MS COCO dataset from scratch. In order to ensure the effectiveness of this approach, numerous detection systems are used in this work. Additionally, positive performance results to reduce the processing time required for object detection have been attained.

PE-Transformer: Path enhanced transformer for improving underwater object detection

Traditional object detection methods cannot effectively identify underwater objects with complex backgrounds, and it is difficult to fully obtain the details of small-scale underwater targets, resulting in poor detection performance. We propose a path-augmented Transformer detection framework to address these limitations to explore the semantic details of small-scale underwater targets in complex environments. On one hand, an embedded local path detection information scheme is devised to facilitate the interaction between high-level and low-level features, thereby enhancing the semantic representation of distinctive features of small-scale underwater targets. Rich dependency relationships are established between the acquired high-level and low-level features within the CSWin-Transformer framework, thus fortifying the semantic representation during the encoding phase. Furthermore, a individualized loss function is employed to optimize and fine-tune features at various hierarchical levels. On the other hand, a detection module with flexible and adaptive point representation that is different from conventional square detection methods is designed. This module covers the underwater target from any direction, and the salient point samples in classification localization and feature selection between points realize the feature selection improves the detection accuracy of underwater objects simultaneously. We designed a new weighted loss function to encourage the network to converge better. Experimental results on open-source underwater and remote sensing images of UTDAC, RUOD, and ADios show that the proposed method outperforms other underwater object detection methods in terms of precision(P), recall(R), comprehensive evaluation index of F1-score and FPS(frames per second).

Real time object recognition based on YOLO model

With the rapid development of computer technology, the concept of computer vision has been proposed. Since then, many object recognition methods have been developed to lay the foundation for computer vision. Object recognition is vital in various computer vision applications, such as autonomous driving, surveillance systems, robotics, and other areas. The You Only Look Once (YOLO) model has gained significant attention due to its ability to achieve real-time object detection and localization in images and videos. This paper comprehensively reviews real-time object recognition based on the YOLO model. We discuss the YOLO architecture's underlying principles and advantages over traditional object detection methods. Then, according to the article by Joseph Redmon, the inventor of YOLO, the benefits of each version of the YOLO model and the performance optimization compared to the previous work are briefly introduced in the order of release. Furthermore, this paper explores its applications in different domains.

Research on efficient detection network method for remote sensing images based on self attention mechanism

Remote sensing images are widely used in aerial drones, satellites, and other fields. However, traditional object detection methods face challenges of low efficiency and accuracy due to the diverse scales, numerous target types, and complex backgrounds of remote sensing images. To address these issues, this paper proposes a remote sensing image detection network based on self-attention, which adaptively learns the relative importance of pixels to achieve more precise and efficient object detection in remote sensing images, meeting the demands of large-scale remote sensing image processing. Firstly, through end-to-end training, the object detection network is optimized as a whole to directly output detection results from the input remote sensing images, eliminating the need for additional intermediate steps. This not only reduces the impact of information loss and inconsistency but also simplifies the entire detection process. Secondly, we integrate the Faster R-CNN architecture as the foundation, combining region extraction and object classification into a unified process. Lastly, we embed self-attention mechanisms at different levels of the Faster R-CNN to progressively extract multi-scale and multi-level feature information, enhancing the network's ability to learn the correlation of information from different positions in the image, automatically capturing the relationships between objects, and improving the accuracy of object detection. This significantly reduces redundant computation, making it more efficient for large-scale remote sensing image processing. Experimental verification demonstrates that this approach outperforms traditional methods in terms of detection accuracy and efficiency, better addressing the particularities of remote sensing images, and providing an efficient and precise solution for aerial drone and satellite image processing. Remote sensing image detection has become one of the research hotspots in the field of remote sensing, holding significant theoretical significance and practical application value.

MSA R-CNN: A comprehensive approach to remote sensing object detection and scene understanding

Remote sensing scene understanding is crucial for extracting valuable information from high-resolution images, including object detection and classification. Traditional object detection methods face challenges in handling the diverse scales, orientations, and complex backgrounds present in remote sensing data. In this paper, we propose a novel remote sensing scene understanding system called multiscale-attention R-CNN (MSA R-CNN), which incorporates a super multiscale feature extraction network (SMENet) for enhanced feature extraction from multiscale images, an adaptive dynamic inner lateral (ADIL) connection module to tackle information loss in feature pyramid networks (FPN), and a distributed lightweight attention module (DLAM) to refine feature information processing. Furthermore, a new dataset combining the DIOR and DOTA datasets is introduced to extract the background information of detected objects and evaluate the proposed system’s performance. MSA R-CNN achieved an mAP of 74.37% on the DIOR dataset when the gamma value was set to 0.2 and 81.97% on the DOTA dataset when the gamma value was set to 0.1 with the same learning rate, outperforming state-of-the-art models on both datasets. The proposed system demonstrates significant improvements in both object detection and background information extraction, providing a comprehensive solution for remote sensing scene understanding.

A Parallel Open-World Object Detection Framework with Uncertainty Mitigation for Campus Monitoring

The recent advancements in artificial intelligence have brought about significant changes in education. In the context of intelligent campus development, target detection technology plays a pivotal role in applications such as campus environment monitoring and the facilitation of classroom behavior surveillance. However, traditional object detection methods face challenges in open and dynamic campus scenarios where unexpected objects and behaviors arise. Open-World Object Detection (OWOD) addresses this issue by enabling detectors to gradually learn and recognize unknown objects. Nevertheless, existing OWOD methods introduce two major uncertainties that limit the detection performance: the unknown discovery uncertainty from the manual generation of pseudo-labels for unknown objects and the known discrimination uncertainty from perturbations that unknown training introduces to the known class features. In this paper, we introduce a Parallel OWOD Framework with Uncertainty Mitigation to alleviate the unknown discovery uncertainty and the known discrimination uncertainty within the OWOD task. To address the unknown discovery uncertainty, we propose an objectness-driven discovery module to focus on capturing the generalized objectness shared among various known classes, driving the framework to discover more potential objects that are distinct from the background, including unknown objects. To mitigate the discrimination uncertainty, we decouple the learning processes for known and unknown classes through a parallel structure to reduce the mutual influence at the feature level and design a collaborative open-world classifier to achieve high-performance collaborative detection of both known and unknown classes. Our framework provides educators with a powerful tool for effective campus monitoring and classroom management. Experimental results on standard benchmarks demonstrate the framework’s superior performance compared to state-of-the-art methods, showcasing its transformative potential in intelligent educational environments.

SA-RPN: A Spacial Aware Region Proposal Network for Acne Detection.

Automated detection of skin lesions offers excellent potential for interpretative diagnosis and precise treatment of acne vulgar. However, the blurry boundary and small size of lesions make it challenging to detect acne lesions with traditional object detection methods. To better understand the acne detection task, we construct a new benchmark dataset named AcneSCU, consisting of 276 facial images with 31777 instance-level annotations from clinical dermatology. To the best of our knowledge, AcneSCU is the first acne dataset with high-resolution imageries, precise annotations, and fine-grained lesion categories, which enables the comprehensive study of acne detection. More importantly, we propose a novel method called Spatial Aware Region Proposal Network(SA-RPN) to improve the proposal quality of two-stage detection methods. Specifically, the representation learning for the classification and localization task is disentangled with a double head component to promote the proposals for hard samples. Then, Normalized Wasserstein Distance of each proposal is predicted to improve the correlation between the classification scores and the proposals' intersection-over-unions(IoUs). SA-RPN can serve as a plug-and-play module to enhance standard two-stage detectors. Extensive experiments are conducted on both AcneSCU and the public dataset ACNE04, and the results show that the proposed method can consistently outperform state-of-the-art methods. Code and the collected dataset are made available at https://github.com/pingguokiller/acnedetection to stimulate the future research in the acne healthcare community.

A Novel Object Detection Method for Solid Waste Incorporating a Weighted Deformable Convolution

Rapid detection of solid waste with remote sensing images is of great significance for environmental protection. In recent years, deep learning-based object detection methods have been widely studied. In contrast to regular objects such as airplanes or buildings, solid wastes commonly h ave arbitrary shapes with difficult‐to‐distinguish boundaries. In this study, a solid waste detection network with a weighted deformable convolution and a global context block based on Feature Pyramid Network (FPN) model was proposed. The designed feature extraction structure can help to enhance the boundary and shape features of solid waste. The effectiveness of the proposed method was verified on the well-known DetectIon in Optical Remote sensing images data set and further on a solid waste data set, which was collected by the authors manually. The experimental results show that the proposed method outperforms other traditional object detection methods and a maximum improvement of 5.27% was obtained compared to the FPN method.

Survey of object detection in terahertz images

In recent years, object detection algorithms have undergone a further development and improvement, resulting in a wider range of application scenarios. As one of the most fundamental and challenging issues in the field of computer vision, the application of object detection in the field of security has also received considerable attention. Terahertz (THz) imaging which is widely used in this area because of the ability to detect hidden objects, as a type of electromagnetic wave imaging with poor imaging performance and low resolution, traditional target detection methods cannot achieve high robustness and effectiveness simultaneously. However, anyway in this possible application scenario, many possible ideas and algorithms have been proposed. This paper analyzes the possibility of applying different object detection methods to terahertz images and analyzes the existing problems in order to give the learner some basic idea and future direction. Several detectors are covered, including the traditional object detection methods and the algorithms based on Convolutional Neural Network(CNN) framework.

Crop disease recognition based on the YOLOv5 and transfer learning

Crop disease detection is an important factor in agricultural production. Traditional object detection methods can't effectively screen key features, resulting in weak crop disease control in many countries. In recent years, several convolutional neural networks for object detection have been proposed, which makes it possible to apply computer vision to crop disease identification through deep learning. YOLOv5 is an advanced object detection network, which can extract key features and use human visual attention mechanism as a reference. This paper would evaluate the performance of four pre-trained models of YOLOv5 in object detection of crop diseases. And transfer learning was used to train the corresponding dataset. The experiment results have showed that the F1 values of the four models all reached above 0.93, and the Yolov5x got the best result, which achieved 0.963. Furthermore, the detection accuracy of the four models has reached more than 98%. This shows that the YOLOv5 series network models have great application prospects in the identification of crop diseases. In the near future, the object detection model can be applied to various mobile devices, even unmanned aerial vehicles, which would play a significant role in crop disease prevention.

HS-YOLO: Small Object Detection for Power Operation Scenarios

Object detection methods are commonly employed in power safety monitoring systems to detect violations in surveillance scenes. However, traditional object detection methods are ineffective for small objects that are similar to the background information in the power monitoring scene, which consequently affects the performance of violation behavior detection. This paper proposed a small object detection algorithm named HS-YOLO, based on High-Resolution Network (HRNet) and sub-pixel convolution. First, to fully extract the microfeature information of the object, a small object feature extraction backbone network is proposed based on the HRNet structure. The feature maps of different scales are processed by multiple parallel branches and fused with each other in the network. Then, to fully retain the effective features of small objects, the sub-pixel convolution module is incorporated as the upsampling operator in the feature fusion network. The low-resolution feature map is upsampled to a higher resolution by reorganizing pixel values and performing padding operations in this module. On our self-constructed power operation dataset, the HS-YOLO algorithm achieved a mAP of 87.2%, which is a 3.5% improvement compared to YOLOv5. Particularly, the dataset’s AP for detecting small objects such as cuffs, necklines, and safety belts is improved by 10.7%, 5.8%, and 4.4%, respectively. These results demonstrate the effectiveness of our proposed method in detecting small objects in power operation scenarios.

RepRCNN: A Structural Reparameterisation Convolutional Neural Network Object Detection Algorithm Based on Branch Matching

A CNN object detection method based on the structural reparameterisation technique using branch matching is proposed to address the problem of balancing accuracy and speed in object detection techniques. By the structural reparameterisation of the convolutional layer in the object detection network, the amount of computation and the number of parameters in the network inference are reduced, the memory overhead is lowered, and the use of the branch-matching method to improve the structural reparameterisation model improves the computational efficiency and speed of the network while maintaining the detection accuracy. Optimisation is also carried out in terms of target screening and loss function, and a new CPC NMS screening strategy was introduced to further improve the performance of the model. The experimental results show that the proposed method achieves competitive results on the PASCAL VOC2012 and MS COCO2017 datasets compared to the traditional object detection methods and the current mainstream models, achieving a better balance between the detection accuracy and detection speed.

Target Detection of Passive Bistatic Radar under the Condition of Impure Reference Signal

This paper addresses the issue of the reference signal being contaminated by target echoes due to factors such as wide beamwidths, beam pointing errors, or site errors in passive bistatic radar (PBR). Such contamination can lead to reference signal impurity and subsequently degrade the performance of traditional object detection methods. To overcome this challenge, a novel target detection method centered around the reference signal’s purification is presented. The proposed method involves the modeling, analysis, localization, and purification of the impure reference signal. First, a new signal model incorporating the impure reference signal is established, followed by a comprehensive analysis of its impact on target detection. Then, leveraging the valuable information hidden in the impure reference signal, target echoes within the impure reference signal are localized and reconstructed. Finally, the reconstructed target echoes are subtracted from the original impure reference signal to purify the reference signal. Based on the purified reference signal, traditional target detection is performed to detect the expected targets. The core advantage of this method lies in its efficient reconstruction and removal of the impure terms within the reference signal. The effectiveness of the proposed method is demonstrated through a series of simulation results.