Object Detection Performance Research Articles

Hierarchical Image Quality Improvement Based on Illumination, Resolution, and Noise Factors for Improving Object Detection

Object detection performance is significantly impacted by image quality factors such as illumination, resolution, and noise. This paper proposes a hierarchical image quality improvement process that dynamically prioritizes these factors based on severity, enhancing detection accuracy in diverse conditions. The process evaluates each factor—illumination, resolution, and noise—using discriminators that analyze brightness, edge strength, and noise levels. Improvements are applied iteratively with an adaptive weight update mechanism that adjusts factor importance based on improvement effectiveness. Following each improvement, a quality assessment is conducted, updating weights to fine-tune subsequent adjustments. This allows the process to learn optimal parameters for varying conditions, enhancing adaptability. The image improved through the proposed process shows improved quality through quality index (PSNR, SSIM) evaluation, and the object detection accuracy is significantly improved when the performance is measured using deep learning models called YOLOv8 and RT-DETR. The detection rate is improved by 7% for the ‘Bottle’ object in a high-light environment, and by 4% and 2.5% for the ‘Bicycle’ and ‘Car’ objects in a low-light environment, respectively. Additionally, segmentation accuracy saw a 9.45% gain, supporting the effectiveness of this method in real-world applications.

Scale-Adaptive Salience Supervision and Dynamic Token Filtering for Small Object Detection in Remote Sensing Images

Abstract Recently, DETR-like detectors, which have shown remarkable performance in general object detection, face limitations when dealing with remote sensing images primarily containing small objects. Mainstream two-stage DETR-like models employ a pipeline that selects and processes a small portion of informative tokens, which enhances performance but also shows a high dependency on token selection. The current static token selection strategies lead to inconsistencies between the static selection criteria and dynamic token updates. Additionally, in remote sensing images, the limited information available for small objects and their inherent sensitivity to pixel shifts further degrade detection performance. To address this, we propose Scale-Adaptive Salience DETR (SAS DETR), a two-stage DETR-like method. SAS DETR incorporates dynamic token filtering, which uses a global threshold predictor to determine the token filtering ratio for each layer of the encoder. This approach selects an appropriate filtering ratio for different network layers while maintaining consistency between the foreground confidence map and token updates. Furthermore, we introduce a novel scale-adaptive salience supervision mechanism that adaptively scales the salience computation area based on object size, ensuring the model more effectively supervises small objects and utilizes the information within tokens without compromising the detection performance for objects of other sizes. Finally, we employ Scale-Adaptive Intersection over Union to reduce the impact of pixel shifts on small objects. With these improvements, our proposed SAS DETR achieves 25.2% AP on the AI-TOD-V2 dataset with 24 training epochs and 50.4% AP on the COCO 2017 dataset with 12 training epochs.

Impact of RAW vs. RGB images on low-light object detection in autonomous driving

Abstract. In the field of autonomous driving, object detection in low-light conditions presents a critical challenge. Traditional detection algorithms, reliant on RGB imagery, often suffer from diminished performance under inadequate illumination. In contrast, RAW images, which retain a higher level of intrinsic image information, have the potential to enhance detection accuracy. Therefore, the paper aims to assess whether models trained on RAW images exhibit enhanced object detection performance in low-light scenarios. To this end, experiments are conducted utilizing the Low-light Object Detection (LOD) dataset, which includes paired RAW and RGB images, with variations in lighting simulated through different exposure times. Two distinct datasets are constructed: one consisting of RAW-normal and RAW-dark images, and another comprising RGB-normal and RGB-dark images. The models evaluated in this research include YOLOv8, Faster R-CNN (Region-based Convolutional Neural Network), and EfficientDet, selected to ensure robustness and generalizability of the findings. The results demonstrate that models trained on RAW images significantly outperform those trained on RGB images, thus exhibiting higher accuracy both in general and under low-light conditions. The preservation of detailed image information in RAW format facilitates enhanced feature extraction, thereby improving detection accuracy and resilience under low-light conditions. These findings provide novel insights into advancing object detection methodologies for autonomous driving systems operating in challenging lighting environments.

YOLOX-S-TKECB: A Holstein Cow Identification Detection Algorithm

Accurate identification of individual cow identity is a prerequisite for the construction of digital farms and serves as the basis for optimized feeding, disease prevention and control, breed improvement, and product quality traceability. Currently, cow identification faces challenges such as poor recognition accuracy, large data volumes, weak model generalization ability, and low recognition speed. Therefore, this paper proposes a cow identification method based on YOLOX-S-TKECB. (1) Based on the characteristics of Holstein cows and their breeding practices, we constructed a real-time acquisition and preprocessing platform for two-dimensional Holstein cow images and built a cow identification model based on YOLOX-S-TKECB. (2) Transfer learning was introduced to improve the convergence speed and generalization ability of the cow identification model. (3) The CBAM attention mechanism module was added to enhance the model’s ability to extract features from cow torso patterns. (4) The alignment between the apriori frame and the target size was improved by optimizing the clustering algorithm and the multi-scale feature fusion method, thereby enhancing the performance of object detection at different scales. The experimental results demonstrate that, compared to the traditional YOLOX-S model, the improved model exhibits a 15.31% increase in mean average precision (mAP) and a 32-frame boost in frames per second (FPS). This validates the feasibility and effectiveness of the proposed YOLOX-S-TKECB-based cow identification algorithm, providing valuable technical support for the application of dairy cow identification in farms.

Using Convolutional Neural Networks for Image Semantic Segmentation and Object Detection

Convolutional neural networks are widely used for feature extraction in the fields of object detection and image segmentation. However, traditional CNN models often struggle to ensure accuracy in high noise environments. A study proposes an enhanced CNN model to improve its ability to recognize targets of different scales. This model combines multi-scale perceptual aggregation and feature alignment (MPAFA) mechanisms. This new method effectively combines low-level and high-level features, which helps to better identify objects of different sizes. The experimental results show that the proposed model achieved a segmentation accuracy of 99.6% on the Cityscapes dataset, and maintained an accuracy of 97.3% even with increased noise. Further experiments have shown that the model outperforms existing methods in terms of accuracy and recall. The experimental results show that the model exhibits excellent performance in object detection and segmentation tasks. This study provides a more effective strategy for processing complex images by optimizing network structure and enhancing feature fusion.

UMS[formula omitted]-ODNet: Unified-scale domain adaptation mechanism driven object detection network with multi-scale attention

Unsupervised domain adaptation techniques improve the generalization capability and performance of detectors, especially when the source and target domains have different distributions. Compared with two-stage detectors, one-stage detectors (especially YOLO series) provide better real-time capabilities and become primary choices in industrial fields. In this paper, to improve cross-domain object detection performance, we propose a Unified-Scale Domain Adaptation Mechanism Driven Object Detection Network with Multi-Scale Attention (UMS2-ODNet). UMS2-ODNet chooses YOLOv6 as the basic framework in terms of its balance between efficiency and accuracy. UMS2-ODNet considers the adaptation consistency across different scale feature maps, which tends to be ignored by existing methods. A unified-scale domain adaptation mechanism is designed to fully utilize and unify the discriminative information from different scales. A multi-scale attention module is constructed to further improve the multi-scale representation ability of features. A novel loss function is created to maintain the consistency of multi-scale information by considering the homology of the descriptions from the same latent feature. Multiply experiments are conducted on four widely used datasets. Our proposed method outperforms other state-of-the-art techniques, illustrating the feasibility and effectiveness of the proposed UMS2-ODNet.

A Study on the Improvement of Object Detection Performance by Infrared Data Augmentation based on Diffusion Models

A Study on the Improvement of Object Detection Performance by Infrared Data Augmentation based on Diffusion Models

Automated Recognition of Submerged Body-like Objects in Sonar Images Using Convolutional Neural Networks

The Police Robot for Inspection and Mapping of Underwater Evidence (PRIME) is an uncrewed surface vehicle (USV) currently being developed for underwater search and recovery teams to assist in crime scene investigation. The USV maps underwater scenes using sidescan sonar (SSS). Test exercises use a clothed mannequin lying on the seafloor as a target object to evaluate system performance. A robust, automated method for detecting human body-shaped objects is required to maximise operational functionality. The use of a convolutional neural network (CNN) for automatic target recognition (ATR) is proposed. SSS image data acquired from four different locations during previous missions were used to build a dataset consisting of two classes, i.e., a binary classification problem. The target object class consisted of 166 196 × 196 pixel image snippets of the underwater mannequin, whereas the non-target class consisted of 13,054 examples. Due to the large class imbalance in the dataset, CNN models were trained with six different imbalance ratios. Two different pre-trained models (ResNet-50 and Xception) were compared, and trained via transfer learning. This paper presents results from the CNNs and details the training methods used. Larger datasets are shown to improve CNN performance despite class imbalance, achieving average F1 scores of 97% in image classification. Average F1 scores for target vs background classification with unseen data are only 47% but the end result is enhanced by combining multiple weak classification results in an ensemble average. The combined output, represented as a georeferenced heatmap, accurately indicates the target object location with a high detection confidence and one false positive of low confidence. The CNN approach shows improved object detection performance when compared to the currently used ATR method.

BGF-YOLOv10: Small Object Detection Algorithm from Unmanned Aerial Vehicle Perspective Based on Improved YOLOv10

With the rapid development of deep learning, unmanned aerial vehicles (UAVs) have acquired intelligent perception capabilities, demonstrating efficient data collection across various fields. In UAV perspective scenarios, captured images often contain small and unevenly distributed objects, and are typically high-resolution. This makes object detection in UAV imagery more challenging compared to conventional detection tasks. To address this issue, we propose a lightweight object detection algorithm, BGF-YOLOv10, specifically designed for small object detection, based on an improved version of YOLOv10n. First, we introduce a novel YOLOv10 architecture tailored for small objects, incorporating BoTNet, variants of C2f and C3 in the backbone, along with an additional small object detection head, to enhance detection performance for small objects. Second, we embed GhostConv into both the backbone and head, effectively reducing the number of parameters by nearly half. Finally, we insert a Patch Expanding Layer module in the neck to restore the feature spatial resolution. Experimental results on the VisDrone-DET2019 and UAVDT datasets demonstrate that our method significantly improves detection accuracy compared to YOLO series networks. Moreover, when compared to other state-of-the-art networks, our approach achieves a substantial reduction in the number of parameters.

YOLO-DroneMS: Multi-Scale Object Detection Network for Unmanned Aerial Vehicle (UAV) Images

In recent years, research on Unmanned Aerial Vehicles (UAVs) has developed rapidly. Compared to traditional remote-sensing images, UAV images exhibit complex backgrounds, high resolution, and large differences in object scales. Therefore, UAV object detection is an essential yet challenging task. This paper proposes a multi-scale object detection network, namely YOLO-DroneMS (You Only Look Once for Drone Multi-Scale Object), for UAV images. Targeting the pivotal connection between the backbone and neck, the Large Separable Kernel Attention (LSKA) mechanism is adopted with the Spatial Pyramid Pooling Factor (SPPF), where weighted processing of multi-scale feature maps is performed to focus more on features. And Attentional Scale Sequence Fusion DySample (ASF-DySample) is introduced to perform attention scale sequence fusion and dynamic upsampling to conserve resources. Then, the faster cross-stage partial network bottleneck with two convolutions (named C2f) in the backbone is optimized using the Inverted Residual Mobile Block and Dilated Reparam Block (iRMB-DRB), which balances the advantages of dynamic global modeling and static local information fusion. This optimization effectively increases the model’s receptive field, enhancing its capability for downstream tasks. By replacing the original CIoU with WIoUv3, the model prioritizes anchoring boxes of superior quality, dynamically adjusting weights to enhance detection performance for small objects. Experimental findings on the VisDrone2019 dataset demonstrate that at an Intersection over Union (IoU) of 0.5, YOLO-DroneMS achieves a 3.6% increase in mAP@50 compared to the YOLOv8n model. Moreover, YOLO-DroneMS exhibits improved detection speed, increasing the number of frames per second (FPS) from 78.7 to 83.3. The enhanced model supports diverse target scales and achieves high recognition rates, making it well-suited for drone-based object detection tasks, particularly in scenarios involving multiple object clusters.

CSSDet: small object detection via cross-scale feature enhancement on drone-view images

ABSTRACT Object detection on drone-view images is vital for applications like intelligent transportation, abnormal behavior detection, and urban surveillance. However, the diverse perspectives and altitudes from which Unmanned Aerial Vehicle (UAV) capture scenes pose challenges due to multi-scale differences and small object sizes. To address this, we propose CSSDet, emphasizing multi-scale object detection and small object enhancement. CSSDet integrates an additional detection head for small objects and employs a bidirectional weighted feature pyramid network for effective cross-scale feature fusion, enhancing global perceptual capability. Additionally, we introduce coordinate attention and Involution modules to mitigate information loss. Experiments are conducted on two publicly available datasets, VisDrone and UAVDT. Quantitative results from the experiments indicate that the proposed CSSDet achieves optimal performance across a wide range of evaluation metrics, with particular strengths demonstrated in multi-scale object detection and small-size object detection. Qualitative results showcase its remarkable effectiveness in various aspects of object detection, including small object detection, objects at different scales and categories, multi-perspective scenarios, and complex scenes. The substantial experimental findings confirm that CSSDet attains comprehensive object detection performance on drone-view images. The source code is available at https://github.com/Gui-Cheng/CSSDet.

Image quality and object detection performance

Image quality and object detection performance

Railway foreign object tracking and detection with spatial positioning and feature generalization enhancement

The existing deep learning foreign object tracking and detection algorithm is easily affected by complex environments and target occlusion, resulting in problems such as missed detection and low detection accuracy. A railway foreign object tracking and detection algorithm with spatial positioning and feature generalization enhancement is proposed. First, a multi-scale cascade improved GhostNet feature network is proposed to improve the feature extraction capability of infrared targets. Secondly, the spatial positioning and feature generalization enhancement module is designed by using the spatial position positioning and generalized morphological information of foreign objects to enhance the detection accuracy of targets with position movement and tracking trajectory changesin complex scenes. Then, a pyramid prediction network is constructed to obtain the detection anchor frame, type and confidence information of infrared railway foreign objects. Finally, by improving the DeepSORT tracking algorithm with improved category and confidence display, combined with Kalman filtering and Hungarian algorithm, railway foreign object tracking and detection in infrared weak light environment is realized. Experimental results show that the proposed method has a tracking and detection accuracy of railway foreign objects 83.3%, with an average detection rate of 11.3 frames per second. Compared with the comparison method, the proposed method has higher detection accuracy and has better performance in railway foreign object tracking and detection in infrared weak light scenes.

Comparing Human Performance on Target Localization in Near Infrared and Long Wave Infrared for Cluttered Environments.

In the context of rapid advancements in AI, the accuracies and speeds among various AI models and methods are often compared. However, a basic question is rarely asked: is AI better than humans, and if so, under what conditions? This paper investigates human ability to detect distant landmark targets under cluttered surroundings such as buildings, trees, and clouds in NIR and LWIR images, aiming to facilitate AI object detection performance analysis. Our investigation employs perception tests and a human performance model to analyze object detection capabilities. The results reveal distinctive differences in NIR and LWIR detectability, showing that although LWIR performs less effectively at range, it offers superior robustness across various environmental conditions. Our findings suggest that AI could be particularly advantageous for object detection in LWIR as it outperform humans in terms of detection accuracy at a long range.

SMALL OBJECT DETECTION APPROACH BASED ON ENHANCED SINGLE-SHOT DETECTOR FOR DETECTION AND RECOGNITION OF INDONESIAN TRAFFIC SIGNS

The detection and recognition of traffic signs are crucial components of advanced driving assistance systems (ADAS) that enhance road safety. Current traffic sign detection and recognition model technology is proficient in identifying and interpreting traffic signs. However, for accurate detection and recognition, the traffic sign in the image must be of a certain minimum pixel size or distance from the driver's sight line for proper detection. The ADAS system should be capable of detecting and recognizing road traffic signs from a considerable distance as they come into the driver's line of vision. The higher the vehicle speed, the greater the distance required for the sign to be detected and recognized, allowing the driver sufficient time to react according to the sign's meaning. Addressing these challenges, this research proposes an enhanced version of the single shot detector (SSD) algorithm, commonly used in object detection, to improve the algorithm's ability to detect small objects. The proposed method involves adding an auxiliary layer module to the original SSD architecture to increase the feature map resolution and expand the conventional layer's receptive space. With the Enhanced SSD algorithm, the detection capability of the SSD can be significantly enhanced in terms of accuracy. The limitations of this study are related to the influence of occlusion and clutter, which might affect the performance of object detection, especially for small objects, which are more susceptible to being influenced by various factors. The research results demonstrate that Enhanced SSD improves object detection accuracy compared to the original SSD, with a mean average precision (mAP) of 97.87 compared to 95.35 for detecting 21 traffic signs in Indonesia.

Evaluation of segmentation and semi‐supervised learning for spotted spurge recognition in bermudagrass turf

AbstractDiverse datasets are crucial for training machine learning‐based weed recognition models. However, annotating (i.e., labeling) images can be laborious and time‐consuming. Choice of annotation method and training approach not only affects the overall model effectiveness but also its minimum training data requirements and development pace. Segmentation and semi‐supervised learning (SSL) may offer performance or training enhancements. This study evaluated (1) segmentation against object detection in spotted spurge [Chamaesyce maculata (L.) Small] recognition in Latitude 36 hybrid bermudagrass [Cynodon dactylon (L.) Pers. × C. transvaalensis Burtt‐Davy] maintained as a golf course fairway and (2) the potential of two‐step SSL‐based training procedure using data labeled both manually and automatically with a pre‐trained model to expedite the model development. The architecture used across this research was You Only Look Once version 8 (YOLOv8) employing nano, small, and medium variants. All models were trained with a dataset restricted to 1200 training and 300 validation images. Both annotation methods resulted in adequate spotted spurge identification, as evidenced with >0.60 (>0.50 acceptability threshold) medium average precision at intersection over union threshold of 0.50 (mAP@50). Although the difference was minimal, object detection performed superior to segmentation. The two‐step training effectively accelerated image annotation while preserving or improving (with 8:4 and 6:6 splits between manually and auto‐labeled data) object detection performance. Segmentation tolerated only 10:2 split and exhibited increased sensitivity to declining dataset quality proportionally to the increase in auto‐labeled images in the final training dataset. Findings show a two‐step SSL‐based training procedure expedites annotation, enhancing model development efficiency.

Multispectral pattern camouflage materials based on PU/Al-flake composite: A CNN-based evaluation

With the continuous progress in optical sensor technologies, the emergence of imagery detection systems operating in the infrared (IR) band has become a reality. In response, we present a novel approach designed to deceive imagery IR detectors targeting fighter jets. To implement this method, we developed a specialized coating technique using a Polyurethane (PU)/Aluminum (Al)-flake composite. This composite coating enables the precise application of IR camouflage patterns while preserving the existing visible camouflages on aircraft surfaces. We customized fighter jet IR image datasets by integrating flow and heat transfer numerical analysis results with real-world background measurements. We then employed the YOLOv3 algorithm, based on a convolutional neural network (CNN), which is known for its high performance in object detection. Through this approach, we analyzed the effects of various pattern shapes on camouflage performance. The proposed PU/Al-flake composite coating offers a versatile and effective solution for overlapping IR camouflage patterns onto existing visible camouflages, providing enhanced protection against advanced IR imagery detection systems. This study highlights the potential of a comprehensive approach to developing and evaluating multispectral camouflage materials and effective camouflage strategies, emphasizing practical applications and real-world performance against IR image detectors.

Pest Detection Based on Lightweight Locality-Aware Faster R-CNN

Accurate and timely monitoring of pests is an effective way to minimize the negative effects of pests in agriculture. Since deep learning-based methods have achieved good performance in object detection, they have been successfully applied for pest detection and monitoring. However, the current pest detection methods fail to balance the relationship between computational cost and model accuracy. Therefore, this paper proposes a lightweight, locality-aware faster R-CNN (LLA-RCNN) method for effective pest detection and real-time monitoring. The proposed model uses MobileNetV3 to replace the original backbone, reduce the computational complexity, and compress the size of the model to speed up pest detection. The coordinate attention (CA) blocks are utilized to enhance the locality information for highlighting the objects under complex backgrounds. Furthermore, the generalized intersection over union (GIoU) loss function and region of interest align (RoI Align) technology are used to improve pest detection accuracy. The experimental results on different types of datasets validate that the proposed model not only significantly reduces the number of parameters and floating-point operations (FLOPs), but also achieves better performance than some popular pest detection methods. This demonstrates strong generalization capabilities and provides a feasible method for pest detection on resource-constrained devices.

Beyond clean data: Exploring the effects of label noise on object detection performance

In recent years, the growth of large-scale datasets has significantly propelled the progress of deep learning applications. Yet, annotating these datasets remains a labor-intensive endeavor, pushing the reliance on cost-effective but less specialized data collection methods and internet data sources. This often results in noisy and inaccurate labels, compromising data quality. Traditional machine learning models assume clean data, but real-world datasets often exhibit significant label noise. This paper examines the impact of such noise on object detection performance, a pivotal aspect of computer vision. We analyze the influence of noisy labels using three renowned object detection frameworks: YOLOv5, Faster R-CNN, and the recent YOLOv8, alongside established datasets: MS COCO, VOC, and ExDARK. Additionally, experiments with the UVM dataset explore domain-specific tasks in dense object scenarios. Two new metrics — Model Health and Detection Capability — were introduced to evaluate the results. Findings indicate that models maintain over 80% of their health (a 20% decline in mAP from the baseline) with up to 40% label corruption. However, Detection Capability deteriorates more sharply under the same conditions. The research also employs the D-RISE method for model explainability, highlighting crucial image regions affecting detection outcomes. Despite the noise, critical detection areas in models remain similar to those in clean data up to the 40% corruption level, as verified by similarity metrics. This study underscores the resilience of object detection models to label noise and provides insights into maintaining performance amidst data quality challenges.

Text-Guided Multi-Class Multi-Object Tracking for Fine-Grained Maritime Rescue

The rapid development of remote sensing technology has provided new sources of data for marine rescue and has made it possible to find and track survivors. Due to the requirement of tracking multiple survivors at the same time, multi-object tracking (MOT) has become the key subtask of marine rescue. However, there exists a significant gap between fine-grained objects in realistic marine rescue remote sensing data and the fine-grained object tracking capability of existing MOT technologies, which mainly focuses on coarse-grained object scenarios and fails to track fine-grained instances. Such a gap limits the practical application of MOT in realistic marine rescue remote sensing data, especially when rescue forces are limited. Given the promising fine-grained classification performance of recent text-guided methods, we delve into leveraging labels and attributes to narrow the gap between MOT and fine-grained maritime rescue. We propose a text-guided multi-class multi-object tracking (TG-MCMOT) method. To handle the problem raised by fine-grained classes, we design a multi-modal encoder by aligning external textual information with visual inputs. We use decoding information at different levels, simultaneously predicting the category, location, and identity embedding features of objects. Meanwhile, to improve the performance of small object detection, we also develop a data augmentation pipeline to generate pseudo-near-infrared images based on RGB images. Extensive experiments demonstrate that our TG-MCMOT not only performs well on typical metrics in the maritime rescue task (SeaDronesSee dataset), but it also effectively tracks open-set categories on the BURST dataset. Specifically, on the SeaDronesSee dataset, the Higher Order Tracking Accuracy (HOTA) reached a score of 58.8, and on the BURST test dataset, the HOTA score for the unknown class improved by 16.07 points.