Imaging Of Small Objects Research Articles

Improved Small Object Detection Algorithm CRL-YOLOv5.

Detecting small objects in images poses significant challenges due to their limited pixel representation and the difficulty in extracting sufficient features, often leading to missed or false detections. To address these challenges and enhance detection accuracy, this paper presents an improved small object detection algorithm, CRL-YOLOv5. The proposed approach integrates the Convolutional Block Attention Module (CBAM) attention mechanism into the C3 module of the backbone network, which enhances the localization accuracy of small objects. Additionally, the Receptive Field Block (RFB) module is introduced to expand the model's receptive field, thereby fully leveraging contextual information. Furthermore, the network architecture is restructured to include an additional detection layer specifically for small objects, allowing for deeper feature extraction from shallow layers. When tested on the VisDrone2019 small object dataset, CRL-YOLOv5 achieved an mAP50 of 39.2%, representing a 5.4% improvement over the original YOLOv5, effectively boosting the detection precision for small objects in images.

SOD-YOLOv8-Enhancing YOLOv8 for Small Object Detection in Aerial Imagery and Traffic Scenes.

Object detection, as a crucial aspect of computer vision, plays a vital role in traffic management, emergency response, autonomous vehicles, and smart cities. Despite the significant advancements in object detection, detecting small objects in images captured by high-altitude cameras remains challenging, due to factors such as object size, distance from the camera, varied shapes, and cluttered backgrounds. To address these challenges, we propose small object detection YOLOv8 (SOD-YOLOv8), a novel model specifically designed for scenarios involving numerous small objects. Inspired by efficient generalized feature pyramid networks (GFPNs), we enhance multi-path fusion within YOLOv8 to integrate features across different levels, preserving details from shallower layers and improving small object detection accuracy. Additionally, we introduce a fourth detection layer to effectively utilize high-resolution spatial information. The efficient multi-scale attention module (EMA) in the C2f-EMA module further enhances feature extraction by redistributing weights and prioritizing relevant features. We introduce powerful-IoU (PIoU) as a replacement for CIoU, focusing on moderate quality anchor boxes and adding a penalty based on differences between predicted and ground truth bounding box corners. This approach simplifies calculations, speeds up convergence, and enhances detection accuracy. SOD-YOLOv8 significantly improves small object detection, surpassing widely used models across various metrics, without substantially increasing the computational cost or latency compared to YOLOv8s. Specifically, it increased recall from 40.1% to 43.9%, precision from 51.2% to 53.9%, mAP0.5 from 40.6% to 45.1%, and mAP0.5:0.95 from 24% to 26.6%. Furthermore, experiments conducted in dynamic real-world traffic scenes illustrated SOD-YOLOv8's significant enhancements across diverse environmental conditions, highlighting its reliability and effective object detection capabilities in challenging scenarios.

Compatibility Review for Object Detection Enhancement through Super-Resolution.

With the introduction of deep learning, a significant amount of research has been conducted in the field of computer vision in the past decade. In particular, research on object detection (OD) continues to progress rapidly. However, despite these advances, some limitations need to be overcome to enable real-world applications of deep learning-based OD models. One such limitation is inaccurate OD when image quality is poor or a target object is small. The performance degradation phenomenon for small objects is similar to the fundamental limitations of an OD model, such as the constraint of the receptive field, which is a difficult problem to solve using only an OD model. Therefore, OD performance can be hindered by low image quality or small target objects. To address this issue, this study investigates the compatibility of super-resolution (SR) and OD techniques to improve detection, particularly for small objects. We analyze the combination of SR and OD models, classifying them based on architectural characteristics. The experimental results show a substantial improvement when integrating OD detectors with SR models. Overall, it was demonstrated that, when the evaluation metrics (PSNR, SSIM) of the SR models are high, the performance in OD is correspondingly high as well. Especially, evaluations on the MS COCO dataset reveal that the enhancement rate for small objects is 9.4% higher compared to all objects. This work provides an analysis of SR and OD model compatibility, demonstrating the potential benefits of their synergistic combination. The experimental code can be found on our GitHub repository.

A novel small object detection algorithm for UAVs based on YOLOv5

Due to the advances in deep learning, artificial intelligence is widely utilized in numerous areas. Technologies frontier, including computer vision, represented by object detection, have endowed unmanned aerial vehicles (UAVs) with autonomous perception, analysis, and decision-making capabilities. UAVs extensively used in numerous fields including photography, industry and agriculture, surveillance, disaster relief, and play an important role in real life. However, current object detection algorithms encountered challenges when it came to detecting small objects in images captured by UAVs. The small size of the objects, with high density, low resolution, and few features make it difficult for the algorithms to achieve high detection accuracy and are prone to miss and false detections especially when detecting small objects. For the case of enhancing the performance of UAV detection on small objects, a novel small object detection algorithm for UAVs adaptation based on YOLOv5s (UA-YOLOv5s) was proposed. (1) To achieve effective small-sized objects detection, a more accurate small object detection (MASOD) structure was adopted. (2) To boost the detection accuracy and generalization ability of the model, a multi-scale feature fusion (MSF) approach was proposed, which fused the feature information of the shallow layers of the backbone and the neck. (3) To enhance the model stability properties and feature extraction capability, a more efficient and stable convolution residual Squeeze-and-Excitation (CRS)module was introduced. Compared with the YOLOv5s, mAP@0.5 was achieved an impressive improvement of 7.2%. Compared with the YOLOv5l, mAP@0.5 increased by 1.0%, and GFLOPs decreased by 69.1%. Compared to the YOLOv3, mAP@0.5 decreased by 0.2% and GFLOPs by 78.5%. The study’s findings demonstrated that the proposed UA-YOLOv5s significantly enhanced the object detection performance of UAVs campared to the traditional algorithms.

Using Improved YOLOv5 Model to Detect Volume for Logs in Log Farms

<p>In this paper, we propose a new computer vision model called SE-YOLOv5-SPD for counting the number of log ends in large wood piles in log farms. This task traditionally requires a lot of manpower and previous computer vision methods struggle to detect logs in low pixels and small objects in images. Our model is based on the YOLOv5 model and incorporates the Squeeze-and-Excitation Networks (SENet) attention module and SPD-Conv (Space-to-Depth Convolution) module to improve accuracy. We also compare the performance of the SE attention module and SPD-Conv module to the CBAM attention module and Focus module using the SE-YOLOv5-SPD model. Results show that the SE-YOLOv5-SPD model can achieve excellent results of 0.652 in mAP50:95, 0.912 in mAP50, 0.968 in Precision, and 0.864 in Recall in a low-resolution environment with interference, which is significantly better than other models. Our findings indicate that the SE-YOLOv5-SPD model is a promising solution for counting the number of log ends in wood piles.</p> <p> </p>

BiTNet: A lightweight object detection network for real-time classroom behavior recognition with transformer and bi-directional pyramid network

In this paper, we propose BiTNet, a real-time object detection network, for enhancing teaching quality and providing feedback through real-time analysis of student behavior in classroom. BiTNet addresses challenges faced by current methods, such as occlusion and small objects in images. Specifically, it incorporates an Efficient Transformer Block (ETB) and Efficient Convolution Aggregation Block (ECAB) for extracting image features. ETB uses convolutional multi-head self-attention (CMHSA) to capture contextual information thus improves the capability to recoginze occluded objects. Its homogeneous multi-branch design reduces the computational costs. The single aggregation operation of ECAB aggregates the extracted features into a single feature map, improving the accuracy of recognition while reducing the computational cost. Then, the semantic and positional information extracted from the backbone are fused in Bi-directional Feature Pyramid Network (BiFPN). BiTNet also employs a smaller size detection head for recognition of small objects. Experimental results on classroom behavior dataset demonstrate that our proposed method has impressive accuracy (82.9%) and speed (256.7 FPS on GPU, 30.7 FPS on CPU, and 18.9 on camera). We further experiment with our method on Pascal VOC 2007 and VisDrone2021, validating BiTNet’s generalization and its capability to recognize small objects. In conclusion, BiTNet has a simple yet efficient network structure and feature extraction module, which reduce computational cost while maintaining recognition accuracy, making it a contender for real-time classroom behavior recognition tasks.

Improving small object detection via cross-layer attention

Small object detection is a fundamental and challenging topic in the computer vision community. To detect small objects in images, several methods rely on feature pyramid networks (FPN), which can alleviate the conflict between resolution and semantic information. However, the FPN-based methods also have limitations. First, existing methods only focus only on regions with close spatial distance, hindering the effectiveness of long-range interactions. Second, element-wise addition ignores the different perceptive fields of the two feature layers, thus causing higher-level features to introduce noise to the lower-level features. To address these problems, we propose a cross-layer attention (CLA) block as a generic block for capturing long-range dependencies and reducing noise from high-level features. Specifically, the CLA block performs feature fusion by factoring in both the channel and spatial dimensions, which provides a reliable way of fusing the features from different layers. Because CLA is a lightweight and general block, it can be plugged into most feature fusion frameworks. On the COCO 2017 dataset, we validated the CLA block by plugging it into several state-of-the-art FPN-based detectors. Experiments show that our approach achieves consistent improvements in both object detection and instance segmentation, which demonstrates the effectiveness of our approach.

Using Deep Neural Networks to Evaluate Leafminer Fly Attacks on Tomato Plants

Among the most common and serious tomato plant pests, leafminer flies (Liriomyza sativae) are considered one of the major tomato-plant-damaging pests worldwide. Detecting the infestation and quantifying the severity of these pests are essential for reducing their outbreaks through effective management and ensuring successful tomato production. Traditionally, detection and quantification are performed manually in the field. This is time-consuming and leads to inaccurate plant protection management practices owing to the subjectivity of the evaluation process. Therefore, the objective of this study was to develop a machine learning model for the detection and automatic estimation of the severity of tomato leaf symptoms of leafminer fly attacks. The dataset used in the present study comprised images of pest symptoms on tomato leaves acquired under field conditions. Manual annotation was performed to classify the acquired images into three groups: background, tomato leaf, and leaf symptoms from leafminer flies. Three models and four different backbones were compared for a multiclass semantic segmentation task using accuracy, precision, recall, and intersection over union metrics. A comparison of the segmentation results revealed that the U-Net model with the Inceptionv3 backbone achieved the best results. For estimation of symptom severity, the best model was FPN with the ResNet34 and DenseNet121 backbones, which exhibited lower root mean square error values. The computational models used proved promising mainly because of their capacity to automatically segment small objects in images captured in the field under challenging lighting conditions and with complex backgrounds.

Development, research of an algorithm for detecting small objects based on convolutional neural networks

The unsatisfactory accuracy of detecting small objects in images of convolutional neural networks is associated with a lack of feature information characterizing small objects, which increases the likelihood of neural network retraining. To increase the feature space, information about the movement of dynamic objects is used. Movement identification of moving objects is based on the background subtraction method. Experimental studies show the superiority of the developed algorithm in the accuracy of detecting small objects in images compared to the original architecture of the convolutional neural network.

A spatial distance-based spatial clustering algorithm for sparse image data

By allocating each object to one of the predefined categories, image classification deeply understands the attributes and features of the data on each object in the scene, and further mines the potential features and internal connections of the data, supporting the subsequent application decision-making with necessary structured data. One of the key challenges to image classification is how to accurately classify sparse data, when there is an imbalance between different categories of data, i.e., how to identify small objects in images. Recognizing a person in satellite images is such a challenging task. These objects are sparse either globally or in each recognizable local segment. Therefore, they are often overlooked by the classifier, or removed as noises. During deep learning, feature sparsity means the samples contain too much useless information, which suppresses the generalization and accuracy of the model. To solve the problem, this paper presents a spatial distance-based spatial clustering algorithm for sparse image data (SDBSCA-SID). Firstly, the imaging range of the image sensor constitutes a two-dimensional (2D) constraint space. Under the constraint, spatial clustering was carried out based on the features of each sample to aggregate dense data into primary categories, and aggregate sparse data and noises into secondary categories. Referring to the 2D constrained space, multiple spatial classification surfaces were constructed to aggregate the sparse data to the two sides of these surfaces as much as possible. If the error is minimized, then the sparse data belong to these classification surfaces. To shorten the convergence time of the clustering algorithm on imbalanced data, the original sample set was cut into slices, and assigned to several calculation units for separate clustering. Next, the same-class clusters were merged through reduction. Finally, the obtained class labels were compared with the preset class labels, wrapping up the semantic segmentation of images. The stability and accuracy of our algorithm were demonstrated through tests on image samples.

Subdiffraction Quantum Imaging with Undetected Photons.

We propose a nonlinear imaging scheme with undetected photons that overcomes the diffraction limit by transferring near-field information at one wavelength to far-field information of a correlated photon with a different wavelength generated through spontaneous photon-pair generation. At the same time, this scheme allows for retrieval of high-contrast images with zero background, making it a highly sensitive scheme for imaging of small objects at challenging spectral ranges with subdiffraction resolutions.

A Study on Classification and Detection of Small Moths Using CNN Model

Currently, there are many limitations to classify images of small objects. In addition, there are limitations such as error detection due to external factors, and there is also a disadvantage that it is difficult to accurately distinguish between various objects. This paper uses a convolutional neural network (CNN) algorithm to recognize and classify object images of very small moths and obtain precise data images. A convolution neural network algorithm is used for image data classification, and the classified image is transformed into image data to learn the topological structure of the image. To improve the accuracy of the image classification and reduce the loss rate, a parameter for finding a fast-optimal point of image classification is set by a convolutional neural network and a pixel image as a preprocessor. As a result of this study, we applied a convolution neural network algorithm to classify the images of very small moths by capturing precise images of the moths. Experimental results showed that the accuracy of classification of very small moths was more than 90%.

Localizing hot spots in Poisson radiation data matrices: nonnegative tensor factorization and phase congruency

Detecting and delineating hot spots in data from radiation sensors is required in applications ranging from monitoring large geospatial areas to imaging small objects in close proximity. This paper describes a computational method for localizing potential hot spots in matrices of independent Poisson data where, in numerical terms, a hot spot is a cluster of locally higher sample mean values (higher Poisson intensity) embedded in lower sample mean values (lower background intensity). Two numerical algorithms are computed sequentially for a 3D array of 2D matrices of gross Poisson counts: (1) nonnegative tensor factorization of the 3D array to maximize a Poisson likelihood and (2) phase congruency in pertinent matrices. The indicators of potential hot spots are closed contours in phase congruency in these matrices. The method is illustrated for simulated Poisson radiation datasets, including visualization of the phase congruency contours. The method may be useful in other applications in which there are matrices of nonnegative counts, provided that a Poisson distribution fits the dataset.

Experimental demonstration of a three-dimensional acoustic hyperlens for super-resolution imaging

Acoustic hyperlenses have recently attracted much attention for promising applications in various fields. Yet the experimental realization of an acoustic hyperlens working in a real three-dimensional (3D) world is still lacking. Here, we theoretically propose and experimentally demonstrate a 3D acoustic hyperlens capable of producing super-resolution imaging for broadband airborne sound. A simple nonresonant metamaterial is designed as a practical implementation that simultaneously ensures tessellation of the curved surface and deep-subwavelength resolution. We analyze the dispersion relationship of the designed metamaterial that converts the evanescent waves into radially propagating modes based on positive extreme anisotropy. The effectiveness of our mechanism is demonstrated both numerically and experimentally via the production of 3D magnifying super-resolution imaging of small objects containing subwavelength patterns within a broad frequency range. We envision the realization of a 3D acoustic hyperlens to offer possibilities for the design of acoustic super-resolution imaging devices and their application in diverse scenarios ranging from medical ultrasound imaging to noninvasive evaluation.

Object Detection in Drone Imagery via Sample Balance Strategies and Local Feature Enhancement

With the advent of drones, new potential applications have emerged for the unconstrained analysis of images and videos from aerial view cameras. Despite the tremendous success of the generic object detection methods developed using ground-based photos, a considerable performance drop is observed when these same methods are directly applied to images captured by Unmanned Aerial Vehicles (UAVs). Usually, most of the work goes into improving the performance of the detector in aspects such as design loss, training sample selection, feature enhancement, and so forth. This paper proposes a detection framework based on an anchor-free detector with several modules, including a sample balance strategies module and super-resolved generated feature module, to improve performance. We proposed the sample balance strategies module to optimize the imbalance among training samples, especially the imbalance between positive and negative, and easy and hard samples. Due to the high frequencies and noisy representation of the small objects in images captured by drones, the detection task is extraordinarily challenging. However, when compared with other algorithms of this kind, our method achieves better results. We also propose a super-resolved generated GAN (Generative Adversarial Network) module with center-ness weights to effectively enhance the local feature map. Finally, we demonstrate our method’s effectiveness with the proposed modules by carrying out a state-of-the-art performance on Visdrone2020 benchmarks.

Centering of a Miniature Rotation Robot for Multi-Directional Imaging Under Microscopy

Viewing a particular region of an object with high magnification is essentially significant in both fundamental research and practical applications, however, imaging the sample from multi-direction is still very challenging for current microscopes. Herein, a centered miniature rotation robot under microscopy is presented to achieve multi-directional imaging of small objects. First, the miniature rotatable robot is constructed by three nano motors, including one rotational and two linear-motors. Then, a new centering strategy of neighborhood-window difference is proposed to center the micro/nano sample onto the rotation axis of the robot (error less than 20 pixels). In experiments, the sample is tested to be imaged from multi-direction (360°) by rotating the robot continuously. To highlight the versatility of this system, we demonstrate the multi-directional imaging under both optical microscope (OM) and scanning electron microscope (SEM). Compared with current techniques, this system is able to provide richer morphology information at small scale benefiting from the multi-directional imaging, which is expected to have long-term impacts on microscopy observation, micro defect inspection, nanomaterial characterization and other related fields.

Ring-array photoacoustic tomography for imaging human finger vasculature.

.For early diagnosis of rheumatoid arthritis (RA), it is important to visualize its potential marker, vascularization in the synovial membrane of the finger joints. Photoacoustic (PA) imaging, which can image blood vessels at high contrast and resolution, is expected to be a potential modality for earlier diagnosis of RA. In previous studies of PA finger imaging, different acoustic schemes, such as linear-shaped arrays, have been utilized, but these have limited detection views, rendering inaccurate reconstruction, and most of them require rotational detection. We are developing a PA system for finger vascular imaging using a ring-shaped array ultrasound (US) transducer. By designing the ring-array sensor based on simulations, using phantom experiments, it was demonstrated that we have created a system that can image small objects around 0.1 to 0.5 mm in diameter. The full width at half maximum of the slice direction of the system was within 2 mm and corresponded to that of the simulation. Moreover, we could clearly visualize healthy index finger vasculature and the location of the distal interphalangeal and proximal interphalangeal joints by PA and US echo images. In the future, this system could be used as a method for visualizing the three-dimensional vascularization of RA patients’ fingers.

Space-Time Processing of Object Images in Passive Radio Imaging Systems

This paper describes the concept and methods of space-time processing of radio images of small objects in a multiposition radiometer system, which make it possible to increase the performance of this system. The spatial processing is carried out in each overlook cycle of scanning radiometers in order to determine the spatial coordinates of objects and mutual orientation of coordinate systems. The time processing is carried out in a sequence of overlook cycles of radiometers for determining the trajectory of the objects that change mutual positions.

Progress on TSV technology for Medipix3RX chip

The progress of Through Silicon Via (TSV) technology for Medipix3RX chip done at DESY is presented here. The goal of this development is to replace the wire bonds in X-ray detectors with TSVs, in order to reduce the dead area between detectors. We obtained the first working chips assembled together with Si based sensors for X-ray detection. The 3D integration technology, including TSV, Re-distribution layer deposition, bump bonding to the Si sensor and bump bonding to the carrier PCB, was done by Fraunhofer Institute IZM in Berlin. After assembly, the module was successfully tested by recording background radiation and making X-ray images of small objects. The active area of the Medipix3RX chip is 14.1 mm×14.1 mm or 256×256 pixels. During TSV processing, the Medipix3RX chip was thinned from 775 μm original thickness, to 130 μm. The diameter of the vias is 40 μm, and the pitch between the vias is 120 μm. A liner filling approach was used to contact the TSV with the RDL on the backside of the Medipix3RX readout chip.

Dual scan CT image recovery from truncated projections.

There are computerized tomography (CT) scanners available commercially for imaging small objects and they are often categorized as mini-CT X-ray machines. One major limitation of these machines is their inability to scan large objects with good image quality because of the truncation of projection data. An algorithm is proposed in this work which enables such machines to scan large objects while maintaining the quality of the recovered image.