Small Objects Research Articles

Implementation of Image Recognition for Human detection in Underwater Images

Recent advances in deep learning have resolved the challenges of detection of objects underwater. Specialized methods have been developed as a result of the particular characteristics of small, fuzzy objects and heterogeneous noise. The Sample-Weighted Network (SWIPE Net) for small object recognition is one of them, as are frameworks with feature enhancement and anchor refining. Additionally, upgraded versions of the attention processes and YOLOv7 have been released. These advancements help with tracking the effects of clean energy technologies, developing accurate and reliable underwater object detection systems, bridging the communication gap between the deaf and hearing-impaired, and automating the analysis of underwater imagery for the extraction of ecological data.

DSSO-YOLO: A fast detection model for densely stacked small object

Visual detection for densely stacked small object (DSSO) has a wide range of applications in the construction, logistics, and import/export industries. Take the construction industry as an example, intelligent rebar counting, can considerably improve the management efficiency in sales, delivery and inventory management. It can also effectively prevent acts such as supervisory theft. However, current mainstream object detection models are susceptible to complex backgrounds when applying directly to DSSO detection, and have not been optimized for highly similar objects with high regularity of arrangement, resulting in low detection accuracy and efficiency. Therefore, we propose the DSSO-YOLO model, which introduces the coordinate attention mechanism into the C3 module of YOLOv5s to increase the accuracy of detection. And the bounding box loss function CIoU in YOLOv5s is replaced with Focal-EIoU-Loss, in order to accelerate the convergence speed and further improve the regression accuracy. To verify the performance of the algorithm, we constructed the steel products dataset SPDC. According to the test results conducted on SPDC, compared with YOLOv5s, the F1-Score, recall, mAP@0.5 and mAP@0.5:0.95 increased by 2.0%, 2.6%, 2.1% and 1.5%, respectively.

Mass Transfer System of a Large Number of Small Objects Based on Conjunction of Triboelectric Nanogenerators and Photo‐Responsive Interface

AbstractMass transfer technology for large quantities of tiny substances, such as electronic chips and drug particles, plays a crucial role in many industries. This study proposes a transferring system for large quantities of small objects based on the synergism of triboelectric nanogenerators (TENGs) and photo‐responsive dielectric materials. TENG device can provide an output voltage of over 8 kV within a rotation time of 40 ms and the titanium oxide phthalocyanine (TiOPc) photoconductive films are fabricated as the photo‐responsive interface, in order to match the output impedance of TENG. The transfer system enables a maximum adsorption load of 98 mg mm−2, which is enough for handling all common electronics chips/wafers. Under the stimulation of laser light, the conductivity of the TiOPc interface can be increased by two orders of magnitude within 1 ms, which leads to the vanish of the surface potential on the laser spot and the release of patterned or pointed object. This photo‐responsive strategy can achieve both large quantity transfer of tiny objects and highly selective release of the element at designed position. This collaborative mechanism of electrostatic force and photoconductivity provides a different approach for realizing efficient and precise Mass transfer system of semiconductor and chip industry.

Efficient Small-Object Detection in Underwater Images Using the Enhanced YOLOv8 Network

Underwater object detection plays a significant role in marine ecosystem research and marine species conservation. The improvement of related technologies holds practical significance. Although existing object-detection algorithms have achieved an excellent performance on land, they are not satisfactory in underwater scenarios due to two limitations: the underwater objects are often small, densely distributed, and prone to occlusion characteristics, and underwater embedded devices have limited storage and computational capabilities. In this paper, we propose a high-precision, lightweight underwater detector specifically optimizing for underwater scenarios based on the You Only Look Once Version 8 (YOLOv8) model. Firstly, we replace the Darknet-53 backbone of YOLOv8s with FasterNet-T0, reducing model parameters by 22.52%, FLOPS by 23.59%, and model size by 22.73%, achieving model lightweighting. Secondly, we add a Prediction Head for Small Objects, increase the number of channels for high-resolution feature map detection heads, and decrease the number of channels for low-resolution feature map detection heads. This results in a 1.2% improvement in small-object detection accuracy, while the remaining model parameters and memory consumption are nearly unchanged. Thirdly, we use Deformable ConvNets and Coordinate Attention in the neck part to enhance the accuracy in the detection of irregularly shaped and densely occluded small targets. This is achieved by learning convolution offsets from feature maps and emphasizing the regions of interest (RoIs). Our method achieves 52.12% AP on the underwater dataset UTDAC2020, with only 8.5 M parameters, 25.5 B FLOPS, and 17 MB model size. It surpasses the performance of large model YOLOv8l, at 51.69% AP, with 43.6 M parameters, 164.8 B FLOPS, and 84 MB model size. Furthermore, by increasing the input image resolution to 1280 × 1280 pixels, our model achieves 53.18% AP, making it the state-of-the-art (SOTA) model for the UTDAC2020 underwater dataset. Additionally, we achieve 84.4% mAP on the Pascal VOC dataset, with a substantial reduction in model parameters compared to previous, well-established detectors. The experimental results demonstrate that our proposed lightweight method retains effectiveness on underwater datasets and can be generalized to common datasets.

Acoustic Actuators for the Manipulation of Micro/Nanorobots: State-of-the-Art and Future Outlooks.

Compared to other actuating methods, acoustic actuators offer the distinctive capability of the contactless manipulation of small objects, such as microscale and nanoscale robots. Furthermore, they have the ability to penetrate the skin, allowing for the trapping and manipulation of micro/nanorobots that carry therapeutic agents in diverse media. In this review, we summarize the current progress in using acoustic actuators for the manipulation of micro/nanorobots used in various biomedical applications. First, we introduce the actuating method of using acoustic waves to manipulate objects, including the principle of operation and different types of acoustic actuators that are usually employed. Then, applications involving manipulating different types of devices are reviewed, including bubble-based microrobots, bubble-free robots, biohybrid microrobots, and nanorobots. Finally, we discuss the challenges and future perspectives for the development of the field.

Radiative loss of coherence in free electrons: a long-range quantum phenomenon

Quantum physics rules the dynamics of small objects as they interact over microscopic length scales. Nevertheless, quantum correlations involving macroscopic distances can be observed between entangled photons as well as in atomic gases and matter waves at low temperatures. The long-range nature of the electromagnetic coupling between charged particles and extended objects could also trigger quantum phenomena over large distances. Here, we reveal a manifestation of quantum mechanics that involves macroscopic distances and results in a nearly complete depletion of coherence associated with which-way free-electron interference produced by electron–radiation coupling in the presence of distant extended objects. This is a ubiquitous effect that we illustrate through a rigorous theoretical analysis of a two-path electron beam interacting with a semi-infinite metallic plate and find the inter-path coherence to vanish proportionally to the path separation at zero temperature and exponentially at finite temperature. The investigated regime of large distances originates in the coupling of the electron to radiative modes assisted by diffraction at material structures but without any involvement of material excitations. Besides the fundamental interest of this macroscopic quantum phenomenon, our results suggest an approach to measuring the vacuum temperature and nondestructively sensing the presence of distant objects.

ESG-YOLO: A Method for Detecting Male Tassels and Assessing Density of Maize in the Field

The intelligent acquisition of phenotypic information on male tassels is critical for maize growth and yield assessment. In order to realize accurate detection and density assessment of maize male tassels in complex field environments, this study used a UAV to collect images of maize male tassels under different environmental factors in the experimental field and then constructed and formed the ESG-YOLO detection model based on the YOLOv7 model by using GELU as the activation function instead of the original SiLU and by adding a dual ECA attention mechanism and an SPD-Conv module. And then, through the model to identify and detect the male tassel, the model’s average accuracy reached a mean value (mAP) of 93.1%; compared with the YOLOv7 model, its average accuracy mean value (mAP) is 2.3 percentage points higher. Its low-resolution image and small object target detection is excellent, and it can be more intuitive and fast to obtain the maize male tassel density from automatic identification surveys. It provides an effective method for high-precision and high-efficiency identification of maize male tassel phenotypes in the field, and it has certain application value for maize growth potential, yield, and density assessment.

Location and segmentation of important railway signs based on improved segmentation

Purpose With the rapid development of railway-intelligent video technology, scene understanding is becoming more and more important. Semantic segmentation is a major part of scene understanding. There is an urgent need for an algorithm with high accuracy and real-time to meet the current railway requirements for railway identification. In response to this demand, this paper aims to explore a variety of models, accurately locate and segment important railway signs based on the improved SegNeXt algorithm, supplement the railway safety protection system and improve the intelligent level of railway safety protection. Design/methodology/approach This paper studies the performance of existing models on RailSem19 and explores the defects of each model through performance so as to further explore an algorithm model dedicated to railway semantic segmentation. In this paper, the authors explore the optimal solution of SegNeXt model for railway scenes and achieve the purpose of this paper by improving the encoder and decoder structure. Findings This paper proposes an improved SegNeXt algorithm: first, it explores the performance of various models on railways, studies the problems of semantic segmentation on railways and then analyzes the specific problems. On the basis of retaining the original excellent MSCAN encoder of SegNeXt, multiscale information fusion is used to further extract detailed features such as multihead attention and mask, solving the problem of inaccurate segmentation of current objects by the original SegNeXt algorithm. The improved algorithm is of great significance for the segmentation and recognition of railway signs. Research limitations/implications The model constructed in this paper has advantages in the feature segmentation of distant small objects, but it still has the problem of segmentation fracture for the railway, which is not completely segmented. In addition, in the throat area, due to the complexity of the railway, the segmentation results are not accurate. Social implications The identification and segmentation of railway signs based on the improved SegNeXt algorithm in this paper is of great significance for the understanding of existing railway scenes, which can greatly improve the classification and recognition ability of railway small object features and can greatly improve the degree of railway security. Originality/value This article introduces an enhanced version of the SegNeXt algorithm, which aims to improve the accuracy of semantic segmentation on railways. The study begins by investigating the performance of different models in railway scenarios and identifying the challenges associated with semantic segmentation on this particular domain. To address these challenges, the proposed approach builds upon the strong foundation of the original SegNeXt algorithm, leveraging techniques such as multi-scale information fusion, multi-head attention, and masking to extract finer details and enhance feature representation. By doing so, the improved algorithm effectively resolves the issue of inaccurate object segmentation encountered in the original SegNeXt algorithm. This advancement holds significant importance for the accurate recognition and segmentation of railway signage.

FAMeLeS: A multispecies and fully automated method to measure morphological leaf traits

Abstract Leaf shape parameters are of key importance to explain the role of energy balance and water economy in plant species distribution, plant productivity and, more generally, in plant–environment interactions. Yet, leaf shape measurements based on image processing are still challenging due to the high diversity of leaf shapes, colours and sizes leading to the development of time‐consuming methods with a narrow field of applicability, sometimes species‐specific or often limited to a few species. We developed a fully automated method for measuring multiple leaf shape parameters (area, perimeter, length, width, circularity and solidity) based on a large image sampling of leaf diversity (including litter) belonging to 587 species and spread over 232 countries worldwide. To evaluate the accuracy of the method to detect small objects, the sampling particularly targeted Mediterranean ecosystems (32 species and 25,205 leaves), in which small leaves often represent methodological challenges. We compared our approach and found that its mean error in leaf area measurement (+0.46%) was 1.7–148 times lower than four existing methods. It was also the only one capable of detecting and measuring all leaves in the test data set, even variegated and small leaves (less than 1 mm2). Its reliability was extensively checked on the largest and most diversified data set ever used. Our method, accessible to the broader scientific community, was simple, rapid and effective on multiple image file types and on a high diversity of leaf size, shape and colour. As such, our approach allows measurement not only on fresh leaves but also on dry leaves, such as leaf litter or leaves from herbaria. This tolerance to leaf characteristics is crucial to increase large‐scale sampling efforts and paves the way for a standardized multispecies approach to measuring leaf morphological traits for ecological and agricultural studies.

Multispectral Object Detection Based on Multilevel Feature Fusion and Dual Feature Modulation

Multispectral object detection is a crucial technology in remote sensing image processing, particularly in low-light environments. Most current methods extract features at a single scale, resulting in the fusion of invalid features and the failure to detect small objects. To address these issues, we propose a multispectral object detection network based on multilevel feature fusion and dual feature modulation (GMD-YOLO). Firstly, a novel dual-channel CSPDarknet53 network is used to extract deep features from visible-infrared images. This network incorporates a Ghost module, which generates additional feature maps through a series of linear operations, achieving a balance between accuracy and speed. Secondly, the multilevel feature fusion (MLF) module is designed to utilize cross-modal information through the construction of hierarchical residual connections. This approach strengthens the complementarity between different modalities, allowing the network to improve multiscale representation capabilities at a more refined granularity level. Finally, a dual feature modulation (DFM) decoupling head is introduced to enhance small object detection. This decoupled head effectively meets the distinct requirements of classification and localization tasks. GMD-YOLO is validated on three public visible-infrared datasets: DroneVehicle, KAIST, and LLVIP. DroneVehicle and LLVIP achieved mAP@0.5 of 78.0% and 98.0%, outperforming baseline methods by 3.6% and 4.4%, respectively. KAIST exhibited an MR of 7.73% with an FPS of 61.7. Experimental results demonstrated that our method surpasses existing advanced methods and exhibits strong robustness.

Microgel-Crosslinked Thermo-Responsive Hydrogel Actuators with High Mechanical Properties and Rapid Response.

Smart hydrogels responsive to external stimuli are promising for various applications such as soft robotics and smart devices. High mechanical strength and fast response rate are particularly important for the construction of hydrogel actuators. Herein, tough hydrogels with rapid response rates are synthesized using vinyl-functionalized poly(N-isopropylacrylamide) (PNIPAM) microgels as macro-crosslinkers and N-isopropylacrylamide as monomers. The compression strength of the obtained PNIPAM hydrogels is up to 7.13MPa. The response rate of the microgel-crosslinked hydrogels is significantly enhanced compared with conventional chemically crosslinked PNIPAM hydrogels. The mechanical strength and response rate of hydrogels can be adjusted by varying the proportion of monomers and crosslinkers. The lower critical solution temperature (LCST) of the PNIPAM hydrogels could be tuned by copolymerizing with ionic monomer sodium methacrylate. Thermo-responsive bilayer hydrogels are fabricated using PINPAM hydrogels with different LCSTs via a layer-by-layer method. The thermo-responsive fast swelling and shrinking properties of the two layers endow the bilayer hydrogel with anisotropic structures and asymmetric response characteristics, allowing the hydrogel to respond rapidly. The bilayer hydrogels are fabricated into clamps to grab small objects and flowers that mimicked the closure of petals, and it shows great application prospects in the field of actuators.

Lightweight Spatial-Temporal Contextual Aggregation Siamese Network for Unmanned Aerial Vehicle Tracking

Benefiting from the powerful feature extraction capability of deep learning, the Siamese tracker stands out due to its advanced tracking performance. However, constrained by the complex backgrounds of aerial tracking, such as low resolution, occlusion, similar objects, small objects, scale variation, aspect ratio change, deformation and limited computational resources, efficient and accurate aerial tracking is still difficult to realize. In this work, we design a lightweight and efficient adaptive temporal contextual aggregation Siamese network for aerial tracking, which is designed with a parallel atrous module (PAM) and adaptive temporal context aggregation model (ATCAM) to mitigate the above problems. Firstly, by using a series of atrous convolutions with different dilation rates in parallel, the PAM can simultaneously extract and aggregate multi-scale features with spatial contextual information at the same feature map, which effectively improves the ability to cope with changes in target appearance caused by challenges such as aspect ratio change, occlusion, scale variation, etc. Secondly, the ATCAM adaptively introduces temporal contextual information to the target frame through the encoder-decoder structure, which helps the tracker resist interference and recognize the target when it is difficult to extract high-resolution features such as low-resolution, similar objects. Finally, experiments on the UAV20L, UAV123@10fps and DTB70 benchmarks demonstrate the impressive performance of the proposed network running at a high speed of over 75.5 fps on the NVIDIA 3060Ti.

A low-cost underactuated compliant gripper with multifunctional jaws for precision manipulation

Precision manipulation tasks often require the use of compliant grippers to handle small and delicate objects accurately. This paper proposes a novel low-cost underactuated compliant gripper with integrated position and force sensors for precision manipulation. The gripper has a millimeter-scale grasping range and can adapt to objects of varying sizes and shapes. This is made possible by a flexure motion transmission mechanism that can undergo significant deformation, an underactuated seesaw flexure mechanism, and a pair of multifunctional jaws. Analytical models based on the pseudo-rigid-body approach are developed to analyze the kinematic, static, and dynamic characteristics of the gripper design. Finite element simulations are conducted to investigate the static and dynamic performance of the gripper. The gripper prototype is fabricated using 3D printing technology for a series of experimental tests. The results of these experiments demonstrate that the developed gripper has excellent motion tracking capabilities and high motion and force resolutions, enabling precise and accurate manipulation tasks. Furthermore, the gripper is capable of successfully grasping objects of different sizes and shapes through parallel and enveloping grasps.

Development of an Isoline Modeling Method for Extended Small Celestial Objects

This article presents a method for modeling extended small celestial (ESC) objects, which are mainly cometary systems. Special attention was given to the possibility of analyzing their structure and physical properties in line with the modern theories of the Solar System’s formation suggesting a rather complicated evolutionary dynamics. Modeling and investigating the structure of different extended celestial objects advance our understanding of the general evolutionary processes that have taken place in the Solar System because all its objects are evolutionarily related. The isoline modeling (IM) method was tested on the real comet data and proved effective in assessing the activity of the processes that occur as ESC objects move in space. The IM method is particularly useful for studying long-period comets that, in many cases, cross the perihelion only once within a foreseeable period of human existence.

One-Stage Small Object Detection Using Super-Resolved Feature Map for Edge Devices

Despite the achievements of deep neural-network-based object detection, detecting small objects in low-resolution images remains a challenging task due to limited information. A possible solution to alleviate the issue involves integrating super-resolution (SR) techniques into object detectors, particularly enhancing feature maps for small-sized objects. This paper explores the impact of high-resolution super-resolved feature maps generated by SR techniques, especially for a one-stage detector that demonstrates a good compromise between detection accuracy and computational efficiency. Firstly, this paper suggests the integration of an SR module named feature texture transfer (FTT) into the one-stage detector, YOLOv4. Feature maps from the backbone and the neck of vanilla YOLOv4 are combined to build a super-resolved feature map for small-sized object detection. Secondly, it proposes a novel SR module with more impressive performance and slightly lower computation demand than the FTT. The proposed SR module utilizes three input feature maps with different resolutions to generate a super-resolved feature map for small-sized object detection. Lastly, it introduces a simplified version of an SR module that maintains similar performance while using only half the computation of the FTT. This attentively simplified module can be effectively used for real-time embedded systems. Experimental results demonstrate that the proposed approach substantially enhances the detection performance of small-sized objects on two benchmark datasets, including a self-built surveillance dataset and the VisDrone2019 dataset. In addition, this paper employs the proposed approach on an embedded system with a Qualcomm QCS610 and demonstrates its feasibility for real-time operation on edge devices.

Palm Gripper Measurement Device for Post-Stroke Rehabilitation Progressive Tracking

Various devices are used in the medical world to measure grip force. However, there is no well-defined method being used to quantify the distribution of grip forces applied by post-stroke patients. It is important to track the patient's progress in neurofeedback training throughout rehabilitation with quantitative evaluation. A palm gripper measurement device is being developed, equipped with Force Sensing Resistors (FSRs) (RP-S40-ST model) to capture grip force. The device provides valuable insights into rehabilitation progress by assessing the grip force. The analogue value from FSRs is linearly interpolated from the input range to the output range using the map function; 'map(avg_force, 0, 1023, 0, 15)' to scale the 'fsrReading' from the input range of 0 to 1023. The input range is converted into a score 0 to 15 scale of a bar graph to indicate the amount of force measured. The accuracy showed by Pearson’s r shows a correlation trend between the analogue value and length of the bar graph with 0.97651 and 0.98083, respectively. A matrix was plotted for three subjects with different object sizes for device testing which shows the adjusted R2 is 0.955 highest for big objects and the lowest adjusted R2 is 0.63672 for small objects.

Versatile correlation learning for size-robust generalized counting: A new perspective

Generalized counting has recently emerged to count novel-class objects within a query image, leveraging limited exemplars. Although methods based on exemplar-query pairs matching have made impressive progress, they typically rely on a single correlation representation, regardless of the varying sizes of objects, which limits more accurate counting. In this paper, we introduce a novel and conceptually straightforward perspective to guide the design of our correlation mechanism that enhances the effectiveness of counting size-diversity objects. Our new perspective encompasses three key aspects: (1) Small objects typically exhibit features concentrated within limited spatial regions, underscoring the importance of an effective channel-wise correlation mechanism for small object counting. (2) Large objects tend to possess rich spatial semantics, making an effective spatial-wise correlation mechanism crucial for large object counting. (3) Integrating both channel-wise and spatial-wise correlation mechanisms holds the potential to enhance counting accuracy across different object sizes. Building upon the above perspective, firstly, we propose a simple yet effective Dual-level Channel-wise Correlation (DCC) module that utilizes kernel-wise correlation and distinct correlation to encode global-to-local channel-wise relationships, enhancing small objects counting accuracy. Secondly, we develop a 4D-convolution-based Spatial-aware Correlation (4DSC) module to extract local-to-local spatial correlation in 4D space, promoting large objects counting accuracy. Finally, we combine the proposed DCC and 4DSC to realize our Versatile Correlation Module (VCM) to simultaneously process both small and large objects, providing adaptability to object size diversity. Extensive experiments on the FSC-147 dataset and CARPK dataset demonstrate the effectiveness of the proposed methods and the superior performance of our counting model.

Improved small foreign object debris detection network based on YOLOv5

Improved small foreign object debris detection network based on YOLOv5

Detection algorithm for dense small objects in high altitude image

The high altitude images captured by drones often contain small-sized objects, which pose several challenges including significant scale variations and dense clustering. To tackle these challenges in small object detection, this paper presents EM-YOLO (Efficient Switching Network and Multi-Scale Feature Fusion Reinforced YOLO for Small Object Detection), an enhanced algorithm based on the YOLOv5s architecture. Firstly, to enhance the network's adaptability to varying object feature scales, the EM-YOLO redesigns the feature extraction backbone network by using different dilation rates in convolutions within the backbone. This flexible structure allows for improved extraction of object features, enhancing the network's feature extraction capability. Secondly, additional detection layer for small objects and combines it with the dynamic detection head. This combination addresses the issue of small object information loss due to multiple down-sampling operations on the feature maps. Finally, Focal-EIoU Loss is employed to enhance regression accuracy in EM-YOLO. Extensive experiments were conducted on the VisDrone2019-DET dataset. The results of these experiments demonstrate that the improved algorithm achieved a significant increase in performance. Specifically, the algorithm achieves 43.5% and 25.1% on the mAP0.5 and mAP0.5:0.95 metrics, which are 13.9% and 8.9% higher than YOLOv5s algorithm, respectively. Furthermore, comparative experiments with other mainstream algorithms indicate that EM-YOLO outperforms them in the task of small object detection.

Inter-Domain Invariant Cross-Domain Object Detection Using Style and Content Disentanglement for In-Vehicle Images

The accurate detection of relevant vehicles, pedestrians, and other targets on the road plays a crucial role in ensuring the safety of autonomous driving. In recent years, object detectors based on Transformers or CNNs have achieved excellent performance in the fully supervised paradigm. However, when the trained model is directly applied to unfamiliar scenes where the training data and testing data have different distributions statistically, the model’s performance may decrease dramatically. To address this issue, unsupervised domain adaptive object detection methods have been proposed. However, these methods often exhibit decreasing performance when the gap between the source and target domains increases. Previous works mainly focused on utilizing the style gap to reduce the domain gap while ignoring the content gap. To tackle this challenge, we introduce a novel method called IDI-SCD that effectively addresses both the style and content gaps simultaneously. Firstly, the domain gap is reduced by disentangling it into the style gap and content gap, generating corresponding intermediate domains in the meanwhile. Secondly, during training, we focus on one single domain gap at a time to achieve inter-domain invariance. That is, the content gap is tackled while maintaining the style gap, and vice versa. In addition, the style-invariant loss is used to narrow down the style gap, and the mean teacher self-training framework is used to narrow down the content gap. Finally, we introduce a multiscale fusion strategy to enhance the quality of pseudo-labels, which mainly focus on enhancing the detection performance for extreme-scale objects (very large or very small objects). We conduct extensive experiments on four mainstream datasets of in-vehicle images. The experimental results demonstrate the effectiveness of our method and its superiority over most of the existing methods.