Manual Annotation Research Articles

Use of High-Frequency Ultrasound for the Correlation of Epithelial Rete Peg Variation

ObjectiveRete pegs are projections of the oral epithelium into connective tissue. Their dimensions change during pathological conditions and may correlate with wound-healing status. Non-invasive, high-frequency ultrasound (US) may be able to capture these changes and aid in early detection of histopathological changes. The aim of this preclinical study is to correlate US images with histology and quantify epithelial layers at different tooth sites. MethodsSagittal B-mode images of mid-facial and interproximal oral soft tissue sites were recorded in a preclinical minipig model using a linear array in second harmonic mode (12/24 MHz). Histology samples from the same locations were stained (hematoxylin and eosin), digitized and registered with US images. Manual annotations were used to measure distances D1 (thickness of epithelium on histology vs. hyperechoic zone on US) and D2 (sum of epithelial thickness and length of rete pegs on histology vs. sum of hyperechoic and hypoechoic zone on US) to statistically analyze them. ResultsUltrasonic-derived dimensions yielded a mean bias of -0.64 (55% coefficient of variance [COV]: -180 to +180 µm) and -12 µm (39% COV: -260 to +240 µm) for D1 and D2, respectively. Individualized analysis of D1 and D2 by tooth type showed similar tends in the ability to differentiate between epithelium at different tooth locations, on both histology and US. ConclusionAssessing soft tissue dimensions on a sub-millimeter scale using clinical imaging hardware is still a developing area. Future research might open doors for diagnosis of oral pathologies and abnormal wound healing, and may limit false-positive indications for biopsies.

GPT is an effective tool for multilingual psychological text analysis

The social and behavioral sciences have been increasingly using automated text analysis to measure psychological constructs in text. We explore whether GPT, the large-language model (LLM) underlying the AI chatbot ChatGPT, can be used as a tool for automated psychological text analysis in several languages. Across 15 datasets (n = 47,925 manually annotated tweets and news headlines), we tested whether different versions of GPT (3.5 Turbo, 4, and 4 Turbo) can accurately detect psychological constructs (sentiment, discrete emotions, offensiveness, and moral foundations) across 12 languages. We found that GPT (r = 0.59 to 0.77) performed much better than English-language dictionary analysis (r = 0.20 to 0.30) at detecting psychological constructs as judged by manual annotators. GPT performed nearly as well as, and sometimes better than, several top-performing fine-tuned machine learning models. Moreover, GPT's performance improved across successive versions of the model, particularly for lesser-spoken languages, and became less expensive. Overall, GPT may be superior to many existing methods of automated text analysis, since it achieves relatively high accuracy across many languages, requires no training data, and is easy to use with simple prompts (e.g., "is this text negative?") and little coding experience. We provide sample code and a video tutorial for analyzing text with the GPT application programming interface. We argue that GPT and other LLMs help democratize automated text analysis by making advanced natural language processing capabilities more accessible, and may help facilitate more cross-linguistic research with understudied languages.

Pavement Crack Detection Using Fractal Dimension and Semi-Supervised Learning

Pavement cracks are crucial indicators for assessing the structural health of asphalt roads. Existing automated crack detection models depend on large quantities of precisely annotated crack sample data. The irregular morphology of cracks makes manual annotation time-consuming and costly, thereby posing challenges to the practical application of these models. This study proposes a pavement crack image detection method integrating fractal dimension analysis and semi-supervised learning. It identifies the self-similarity characteristics within the crack regions by analyzing pavement crack images and using fractal dimensions to preliminarily determine the candidate crack regions. The Crack Similarity Learning Network (CrackSL-Net) is then employed to learn the semantic similarity of crack image regions. Semi-supervised learning facilitates automatic crack detection by combining a small amount of labeled data with a large volume of unlabeled image data. Comparative experiments are conducted on two public pavement crack datasets against the HED, U-Net, and RCF models to comprehensively evaluate the performance of the proposed method. The results indicate that, with a 50% annotation ratio, the proposed method achieves high-precision crack detection, with an intersection over union (IoU) exceeding 0.84, which is close to that of U-Net. Visual analysis of the detection results confirms the method’s effectiveness in identifying cracks in complex environments.

Dual-branch teacher-student with noise-tolerant learning for domain adaptive nighttime segmentation

While significant progress has been achieved in the field of image semantic segmentation, the majority of research has been primarily concentrated on daytime scenes. Semantic segmentation of nighttime images is equally critical for autonomous driving; however, this task presents greater challenges due to inadequate lighting and difficulties associated with obtaining accurate manual annotations. In this paper, we introduce a novel method called the Dual-Branch Teacher-Student (DBTS) framework for unsupervised nighttime semantic segmentation. Our approach combines domain alignment and knowledge distillation in a mutually reinforcing manner. Firstly, we employ a photometric alignment module to dynamically generate target-like latent images, bridging the appearance gap between the source domain (daytime) and the target domain (nighttime). Secondly, we establish a dual-branch framework, where each branch enhances collaboration between the teacher and student networks. The student network utilizes adversarial learning to align the target domain with another domain (i.e., source or latent domain), while the teacher network generates reliable pseudo-labels by distilling knowledge from the latent domain. Furthermore, recognizing the potential noise present in pseudo-labels, we propose a noise-tolerant learning method to mitigate the risks associated with overreliance on pseudo-labels during domain adaptation. When evaluated on benchmark datasets, the proposed DBTS achieves state-of-the-art performance. Specifically, DBTS, using different backbones, outperforms established baseline models by approximately 25% in mIoU on the Zurich dataset and by over 26% in mIoU on the ACDC dataset, demonstrating the effectiveness of our method in addressing the challenges of domain-adaptive nighttime segmentation.

Evaluation on Network Social Media Named Entity Recognition Model Based on Active Learning

The medical security privacy and named entity recognition (NER) technology under the blockchain technology has been a hot topic in all walks of life. As a typical representative of medical security risks and NER, the NER model of online social media based on active learning has attracted worldwide attention. NER is an important part of natural language processing. Traditional recognition technology usually requires a lot of external information, and through the manual identification of its features, which costs a lot of time and energy. In order to solve the shortcomings of traditional recognition algorithms and the lack of feature extraction in network media NER, a new active learning model was introduced in this paper. In the information age, people are increasingly demanding a large amount of text information, and NER technology came into being. Its main function is to accurately identify important information from text and provide useful information for high-level work. The initial design of NER system is mainly based on the recognition of rules, so as to realize the recognition of named entities. However, in a complex network environment, it takes a lot of time and energy to establish rules without conflicts, and it has poor mobility. In recent years, with the continuous development of computer technology, the use of machine learning to actively learn the unknown information in the target area reduces the workload of manual annotation, thus realizing the active learning of large amounts of data. The research showed that the recognition accuracy under the traditional NER was low, and the information processing speed was slow; the accuracy rate of NER based on active learning was as high as 97%, and the speed of information processing had also been greatly improved, which had solved many problems under the traditional mode. User satisfaction could be as high as 95%, which showed that the latter had broad prospects. The progress of the new era cannot be separated from the support of new technologies. The research of this article has important guiding significance for medical security privacy and the application of NER under blockchain technology.

Interactive Cascaded Network for Prostate Cancer Segmentation from Multimodality MRI with Automated Quality Assessment.

The accurate segmentation of prostate cancer (PCa) from multiparametric MRI is crucial in clinical practice for guiding biopsy and treatment planning. Existing automated methods often lack the necessary accuracy and robustness in localizing PCa, whereas interactive segmentation methods, although more accurate, require user intervention on each input image, thereby limiting the cost-effectiveness of the segmentation workflow. Our innovative framework addresses the limitations of current methods by combining a coarse segmentation network, a rejection network, and an interactive deep network known as Segment Anything Model (SAM). The coarse segmentation network automatically generates initial segmentation results, which are evaluated by the rejection network to estimate their quality. Low-quality results are flagged for user interaction, with the user providing a region of interest (ROI) enclosing the lesions, whereas for high-quality results, ROIs were cropped from the automatic segmentation. Both manually and automatically defined ROIs are fed into SAM to produce the final fine segmentation. This approach significantly reduces the annotation burden and achieves substantial improvements by flagging approximately 20% of the images with the lowest quality scores for manual annotation. With only half of the images manually annotated, the final segmentation accuracy is statistically indistinguishable from that achieved using full manual annotation. Although this paper focuses on prostate lesion segmentation from multimodality MRI, the framework can be adapted to other medical image segmentation applications to improve segmentation efficiency while maintaining high accuracy standards.

Weakly supervised semantic segmentation of leukocyte images based on class activation maps.

Leukocytes are an essential component of the human defense system, accurate segmentation of leukocyte images is a crucial step towards automating detection. Most existing methods for leukocyte images segmentation relied on fully supervised semantic segmentation (FSSS) with extensive pixel-level annotations, which are time-consuming and labor-intensive. To address this issue, this paper proposes a weakly supervised semantic segmentation (WSSS) approach for leukocyte images utilizing improved class activation maps (CAMs). Firstly, to alleviate ambiguous boundary problem between leukocytes and background, preprocessing technique is employed to enhance the image quality. Secondly, attention mechanism is added to refine the CAMs generated by improving the matching of local and global features. Random walks, dense conditional random fields and hole filling were leveraged to obtain final pseudo-segmentation labels. Finally, a fully supervised segmentation network is trained with pseudo-segmentation labels. The method is evaluated on BCCD and TMAMD datasets. Experimental results demonstrate that by employing the pseudo segmentation annotations generated through this method can be utilized to train UNet as close as possible to FSSS. This method effectively reduces manual annotation cost while achieving WSSS of leukocyte images.

Progressive expansion for semi-supervised bi-modal salient object detection

Existing bi-modal salient object detection (SOD) methods primarily rely on fully supervised training strategies that require extensive manual annotation. Undoubtedly, extensive manual annotation is time-consuming and laborious, and the fully supervised strategy is also prone to overfitting on the training set. Therefore, we introduce a semi-supervised learning architecture (SSLA) to alleviate these problems while ensuring detection performance. Considering that the inherent training mode and concise architecture of basic SSLA will limit its ability to effectively explore the learning potential of the model, we further propose two optimization strategies, dynamic adjustment and active expansion. Specifically, we dynamically adjust the supervision scheme for unlabeled samples during training so that the model can continuously utilize the model’s gains (pseudo labels) to supervise and guide the model to further explore the unlabeled samples. Furthermore, the active expansion strategy enables the model to acquire more beneficial supervised information and focuses its attention on difficult-to-segment samples. In summary, an effective progressive expansion network (PENet) architecture for semi-supervised bi-modal SOD is proposed. Extensive experiments indicate that our PENet architecture, while effectively alleviating 90% of annotation burdens, has achieved highly competitive results in RGB-T and RGB-D tasks compared to fully supervised methods. The performance is even more pronounced during cross-dataset testing.

Weakly Supervised Lymph Nodes Segmentation Based on Partial Instance Annotations with Pre-trained Dual-branch Network and Pseudo Label Learning

Assessing the presence of potentially malignant lymph nodes aids in estimating cancer progression, and identifying surrounding benign lymph nodes can assist in determining potential metastatic pathways for cancer. For quantitative analysis, automatic segmentation of lymph nodes is crucial. However, due to the labor-intensive and time-consuming manual annotation process required for a large number of lymph nodes, it is more practical to annotate only a subset of the lymph node instances to reduce annotation costs. In this study, we propose a pre-trained Dual-Branch network with Dynamically Mixed Pseudo label (DBDMP) to learn from partial instance annotations for lymph nodes segmentation. To obtain reliable pseudo labels for lymph nodes that are not annotated, we employ a dual-decoder network to generate different outputs that are then dynamically mixed. We integrate the original weak partial annotations with the mixed pseudo labels to supervise the network. To further leverage the extensive amount of unannotated voxels, we apply a self-supervised pre-training strategy to enhance the model’s feature extraction capability. Experiments on the mediastinal Lymph Node Quantification (LNQ) dataset demonstrate that our method, compared to directly learning from partial instance annotations, significantly improves the Dice Similarity Coefficient (DSC) from 11.04% to 54.10% and reduces the Average Symmetric Surface Distance (ASSD) from 20.83 mm to 8.72 mm. The code is available at <a href='https://github.com/WltyBY/LNQ2023_training_code'>https://github.com/WltyBY/LNQ2023_training_code</a>

Utilizing deep learning model for assessing melanocytic density in resection margins of lentigo maligna

BackgroundSurgical excision with clear histopathological margins is the preferred treatment to prevent progression of lentigo maligna (LM) to invasive melanoma. However, the assessment of resection margins on sun-damaged skin is challenging. We developed a deep learning model for detection of melanocytes in resection margins of LM.MethodsIn total, 353 whole slide images (WSIs) were included. 295 WSIs were used for training and 58 for validation and testing. The algorithm was trained with 3,973 manual pixel-wise annotations. The AI analyses were compared to those of three blinded dermatopathologists and two pathology residents, who performed their evaluations without AI and AI-assisted. Immunohistochemistry (SOX10) served as the reference standard. We used a dichotomized cutoff for low and high risk of recurrence (≤ 25 melanocytes in an area of 0.5 mm for low risk and > 25 for high risk).ResultsThe AI model achieved an area under the receiver operating characteristic curve (AUC) of 0.84 in discriminating margins with low and high recurrence risk. In comparison, the AUC for dermatopathologists ranged from 0.72 to 0.90 and for the residents in pathology, 0.68 to 0.80. Additionally, with aid of the AI model the performance of two pathologists significantly improved.ConclusionsThe deep learning showed notable accuracy in detecting resection margins of LM with a high versus low risk of recurrence. Furthermore, the use of AI improved the performance of 2/5 pathologists. This automated tool could aid pathologists in the assessment or pre-screening of LM margins.

Fully Automated Hippocampus Segmentation using T2-informed Deep Convolutional Neural Networks

Hippocampal atrophy (tissue loss) has become a fundamental outcome parameter in clinical trials on Alzheimer's disease. To accurately estimate hippocampus volume and track its volume loss, a robust and reliable segmentation is essential. Manual hippocampus segmentation is considered the gold standard but is extensive, time-consuming, and prone to rater bias. Therefore, it is often replaced by automated programs like FreeSurfer, one of the most commonly used tools in clinical research. Recently, deep learning-based methods have also been successfully applied to hippocampus segmentation. The basis of all approaches are clinically used T1-weighted whole-brain MR images with approximately 1 mm isotropic resolution. However, such T1 images show low contrast-to-noise ratios (CNRs), particularly for many hippocampal substructures, limiting delineation reliability. To overcome these limitations, high-resolution T2-weighted scans are suggested for better visualization and delineation, as they show higher CNRs and usually allow for higher resolutions. Unfortunately, such time-consuming T2-weighted sequences are not feasible in a clinical routine.We propose an automated hippocampus segmentation pipeline leveraging deep learning with T2-weighted MR images for enhanced hippocampus segmentation of clinical T1-weighted images based on a series of 3D convolutional neural networks and a specifically acquired multi-contrast dataset. This dataset consists of corresponding pairs of T1- and high-resolution T2-weighted images, with the T2 images only used to create more accurate manual ground truth annotations and to train the segmentation network. The T2-based ground truth labels were also used to evaluate all experiments by comparing the masks visually and by various quantitative measures.We compared our approach with four established state-of-the-art hippocampus segmentation algorithms (FreeSurfer, ASHS, HippoDeep, HippMapp3r) and demonstrated a superior segmentation performance. Moreover, we found that the automated segmentation of T1-weighted images benefits from the T2-based ground truth data.In conclusion, this work showed the beneficial use of high-resolution, T2-based ground truth data for training an automated, deep learning-based hippocampus segmentation and provides the basis for a reliable estimation of hippocampal atrophy in clinical studies.

Biomedical literature mining: graph kernel-based learning for gene–gene interaction extraction

The supervised machine learning method is often used for biomedical relationship extraction. The disadvantage is that it requires much time and money to manually establish an annotated dataset. Based on distant supervision, the knowledge base is combined with the corpus, thus, the training corpus can be automatically annotated. As many biomedical databases provide knowledge bases for study with a limited number of annotated corpora, this method is practical in biomedicine. The clinical significance of each patient’s genetic makeup can be understood based on the healthcare provider’s genetic database. Unfortunately, the lack of previous biomedical relationship extraction studies focuses on gene–gene interaction. The main purpose of this study is to develop extraction methods for gene–gene interactions that can help explain the heritability of human complex diseases. This study referred to the information on gene–gene interactions in the KEGG PATHWAY database, the abstracts in PubMed were adopted to generate the training sample set, and the graph kernel method was adopted to extract gene–gene interactions. The best assessment result was an F1-score of 0.79. Our developed distant supervision method automatically finds sentences through the corpus without manual labeling for extracting gene–gene interactions, which can effectively reduce the time cost for manual annotation data; moreover, the relationship extraction method based on a graph kernel can be successfully applied to extract gene–gene interactions. In this way, the results of this study are expected to help achieve precision medicine.

QRS detection in single-lead, telehealth electrocardiogram signals: Benchmarking open-source algorithms.

A key step in electrocardiogram (ECG) analysis is the detection of QRS complexes, particularly for arrhythmia detection. Telehealth ECGs present a new challenge for automated analysis as they are noisier than traditional clinical ECGs. The aim of this study was to identify the best-performing open-source QRS detector for use with telehealth ECGs. The performance of 18 open-source QRS detectors was assessed on six datasets. These included four datasets of ECGs collected under supervision, and two datasets of telehealth ECGs collected without clinical supervision. The telehealth ECGs, consisting of single-lead ECGs recorded between the hands, included a novel dataset of 479 ECGs collected in the SAFER study of screening for atrial fibrillation (AF). Performance was assessed against manual annotations. A total of 12 QRS detectors performed well on ECGs collected under clinical supervision (F1 score ≥0.96). However, fewer performed well on telehealth ECGs: five performed well on the TELE ECG Database; six performed well on high-quality SAFER data; and performance was poorer on low-quality SAFER data (three QRS detectors achieved F1 of 0.78-0.84). The presence of AF had little impact on performance. The Neurokit and University of New South Wales QRS detectors performed best in this study. These performed sufficiently well on high-quality telehealth ECGs, but not on low-quality ECGs. This demonstrates the need to handle low-quality ECGs appropriately to ensure only ECGs which can be accurately analysed are used for clinical decision making.

Using diffusion models to generate synthetic labeled data for medical image segmentation.

Medical image analysis has become a prominent area where machine learning has been applied. However, high-quality, publicly available data are limited either due to patient privacy laws or the time and cost required for experts to annotate images. In this retrospective study, we designed and evaluated a pipeline to generate synthetic labeled polyp images for augmenting medical image segmentation models with the aim of reducing this data scarcity. We trained diffusion models on the HyperKvasir dataset, comprising 1000 images of polyps in the human GI tract from 2008 to 2016. Qualitative expert review, Fréchet Inception Distance (FID), and Multi-Scale Structural Similarity (MS-SSIM) were tested for evaluation. Additionally, various segmentation models were trained with the generated data and evaluated using Dice score (DS) and Intersection over Union (IoU). Our pipeline produced images more akin to real polyp images based on FID scores. Segmentation model performance also showed improvements over GAN methods when trained entirely, or partially, with synthetic data, despite requiring less compute for training. Moreover, the improvement persists when tested on different datasets, showcasing the transferability of the generated images. The proposed pipeline produced realistic image and mask pairs which could reduce the need for manual data annotation when performing a machine learning task. We support this use case by showing that the methods proposed in this study enhanced segmentation model performance, as measured by Dice and IoU scores, when trained fully or partially on synthetic data.

AI4LUC: deep learning and automated mask labelling to support land use and land cover mapping in the Cerrado biome

ABSTRACT Land use and land cover (LULC) mapping in the Brazilian Cerrado is challenging due to its vastness, seasonal vegetation changes, and spectral class confusion. Artificial intelligence, particularly deep learning (DL), offers promising solutions through semantic segmentation of Cerrado images, though it traditionally requires manual image annotation. This paper introduces the AI4LUC (Artificial Intelligence for Land Use and Land Cover Classification) method, leveraging DL and automated mask labeling to address LULC in the Brazilian Cerrado. AI4LUC includes pre-processing modules and the Smart Mask©Labelling (SML) module, which combines morphological operations with the CerraNetv3 deep convolutional neural network to automatically generate training masks for DL models like U-Net and DeepLabv3+. Additionally, we present the Cerrado Dataset (CerraData-v3), a high-resolution image dataset with 80,000 labeled samples. Our experiments show the SML module achieved an F1-score of 0.6647 and mIoU of 0.5068, demonstrating significant potential for automated semantic segmentation in this large biome. Furthermore, DeepLabv3+ outperformed U-Net by 69.86%. Source code and dataset: https://github.com/ai4luc/AI4LUC.

Discovery and generalization of tissue structures from spatial omics data

Tissues are organized into anatomical and functional units at different scales. New technologies for high-dimensional molecular profiling in situ have enabled the characterization of structure-function relationships in increasing molecular detail. However, it remains a challenge to consistently identify key functional units across experiments, tissues, and disease contexts, a task that demands extensive manual annotation. Here, we present spatial cellular graph partitioning (SCGP), a flexible method for the unsupervised annotation of tissue structures. We further present a reference-query extension pipeline, SCGP-Extension, that generalizes reference tissue structure labels to previously unseen samples, performing data integration and tissue structure discovery. Our experiments demonstrate reliable, robust partitioning of spatial data in a wide variety of contexts and best-in-class accuracy in identifying expertly annotated structures. Downstream analysis on SCGP-identified tissue structures reveals disease-relevant insights regarding diabetic kidney disease, skin disorder, and neoplastic diseases, underscoring its potential to drive biological insight and discovery from spatial datasets.

Improving quality control of whole slide images by explicit artifact augmentation

The problem of artifacts in whole slide image acquisition, prevalent in both clinical workflows and research-oriented settings, necessitates human intervention and re-scanning. Overcoming this challenge requires developing quality control algorithms, that are hindered by the limited availability of relevant annotated data in histopathology. The manual annotation of ground-truth for artifact detection methods is expensive and time-consuming. This work addresses the issue by proposing a method dedicated to augmenting whole slide images with artifacts. The tool seamlessly generates and blends artifacts from an external library to a given histopathology dataset. The augmented datasets are then utilized to train artifact classification methods. The evaluation shows their usefulness in classification of the artifacts, where they show an improvement from 0.10 to 0.01 AUROC depending on the artifact type. The framework, model, weights, and ground-truth annotations are freely released to facilitate open science and reproducible research.

HiSEG: Human assisted instance segmentation

Instance segmentation is a form of image detection which has a range of applications, such as object refinement, medical image analysis, and image/video editing, all of which demand a high degree of accuracy. However, this precision is often beyond the reach of what even state-of-the-art, fully automated instance segmentation algorithms can deliver. The performance gap becomes particularly prohibitive for small and complex objects. Practitioners typically resort to fully manual annotation, which can be a laborious process. In order to overcome this problem, we propose a novel approach to enable more precise predictions and generate higher-quality segmentation masks for high-curvature, complex and small-scale objects. Our human-assisted segmentation method, HiSEG, augments the existing Strong Mask R-CNN network to incorporate human-specified partial boundaries. We also present a dataset of hand-drawn partial object boundaries, which we refer to as “human attention maps”. In addition, the Partial Sketch Object Boundaries (PSOB) dataset contains hand-drawn partial object boundaries which represent curvatures of an object’s ground truth mask with several pixels. Through extensive evaluation using the PSOB dataset, we show that HiSEG outperforms state-of-the art methods such as Mask R-CNN, Strong Mask R-CNN, Mask2Former, and Segment Anything, achieving respective increases of +42.0, +34.9, +29.9, and +13.4 points in APMask metrics for these four models. We hope that our novel approach will set a baseline for future human-aided deep learning models by combining fully automated and interactive instance segmentation architectures.

Improving pollen-bearing honey bee detection from videos captured at hive entrance by combining deep learning and handling imbalance techniques

The number of pollen-bearing honey bees serves as a vital indicator for assessing colony balance and health. Despite its significance, prevailing detection techniques still rely heavily on manual observation and annotation, leading to time-consuming processes that cannot sustain long-term, continuous monitoring efforts. To facilitate automatic beehive monitoring, this study introduces an efficient method for pollen-bearing bee detection. Initially, we furnish a comprehensive dataset, dubbed VnPollenBee, meticulously annotated for pollen-bearing honey bee detection and classification. The dataset comprises 60,826 annotated boxes that delineate both pollen-bearing and non-pollen-bearing bees in 2051 images captured at the entrances of beehives under various environmental conditions. To the best of our knowledge, this represents the first dedicated dataset for pollen-bearing bee detection. The VnPollenBee dataset is publicly accessible to the research community at https://comvis-hust.github.io/datasets/pollenbee.html. Subsequently, we propose the incorporation of diverse techniques into two baseline models, namely YOLOv5 and Faster RCNN, to effectively address the imbalance that arises during the detection of pollen-bearing bees due to their number being typically much lower than the total number of bees present at hive entrances. The experimental results demonstrate that our proposed method outperforms the baseline models on the VnPollenBee dataset, yielding Precision, Recall, and F1 score of 99%, 93%, and 95%, respectively. Specifically, the improvements obtained are 3% and 2% in Recall and F1 score when using YOLOv5, and 3%, 2%, and 2% in Precision, Recall, and F1 score when using Faster RCNN. These findings confirm the potential of our approach to facilitate bee foraging behavior analysis and automated bee monitoring.

A Self-Supervised Few-Shot Semantic Segmentation Method Based on Multi-Task Learning and Dense Attention Computation.

Nowadays, autonomous driving technology has become widely prevalent. The intelligent vehicles have been equipped with various sensors (e.g., vision sensors, LiDAR, depth cameras etc.). Among them, the vision systems with tailored semantic segmentation and perception algorithms play critical roles in scene understanding. However, the traditional supervised semantic segmentation needs a large number of pixel-level manual annotations to complete model training. Although few-shot methods reduce the annotation work to some extent, they are still labor intensive. In this paper, a self-supervised few-shot semantic segmentation method based on Multi-task Learning and Dense Attention Computation (dubbed MLDAC) is proposed. The salient part of an image is split into two parts; one of them serves as the support mask for few-shot segmentation, while cross-entropy losses are calculated between the other part and the entire region with the predicted results separately as multi-task learning so as to improve the model's generalization ability. Swin Transformer is used as our backbone to extract feature maps at different scales. These feature maps are then input to multiple levels of dense attention computation blocks to enhance pixel-level correspondence. The final prediction results are obtained through inter-scale mixing and feature skip connection. The experimental results indicate that MLDAC obtains 55.1% and 26.8% one-shot mIoU self-supervised few-shot segmentation on the PASCAL-5i and COCO-20i datasets, respectively. In addition, it achieves 78.1% on the FSS-1000 few-shot dataset, proving its efficacy.