High Annotation Cost Research Articles

OBCTeacher: Resisting labeled data scarcity in oracle bone character detection by semi-supervised learning

Oracle bone characters (OBCs) are ancient ideographs for divination and memorization, as well as first-hand evidence of ancient Chinese culture. The detection of OBC is the premise of advanced studies and was mainly done by authoritative experts in the past. Deep learning techniques have great potential to facilitate OBC detection, but the high annotation cost of OBC brings the scarcity of labeled data, hindering its application. This paper proposes a novel OBC detection framework called OBCTeacher based on semi-supervised learning (SSL) to resist labeled data scarcity. We first construct a large-scale OBC detection dataset. Through investigation, we find that spatial mismatching and class imbalance problems lead to decreased positive anchors and biased predictions, affecting the quality of pseudo labels and the performance of OBC detection. To mitigate the spatial mismatching problem, we introduce a geometric-priori-based anchor assignment strategy and a heatmap polishing procedure to increase positive anchors and improve the quality of pseudo labels. As for the class imbalance problem, we propose a re-weighting method based on estimated class information and a contrastive anchor loss to achieve prioritized learning on different OBC classes and better class boundaries. We evaluate our method by using only a small portion of labeled data while using the remaining data as unlabeled and all labeled data with extra unlabeled data. The results demonstrate the effectiveness of our method compared with other state-of-the-art methods by superior performance and significant improvements of an average of 11.97 in AP50:95 against the only supervised baseline. In addition, our method achieves comparable performance using only 20% of labeled data to the fully-supervised baseline using 100% of labeled data, demonstrating that our method significantly reduces the dependence on labeled data for OBC detection.

Single-Image-Based Deep Learning for Precise Atomic Defect Identification.

Defect engineering is widely used to impart the desired functionalities on materials. Despite the widespread application of atomic-resolution scanning transmission electron microscopy (STEM), traditional methods for defect analysis are highly sensitive to random noise and human bias. While deep learning (DL) presents a viable alternative, it requires extensive amounts of training data with labeled ground truth. Herein, employing cycle generative adversarial networks (CycleGAN) and U-Nets, we propose a method based on a single experimental STEM image to tackle high annotation costs and image noise for defect detection. Not only atomic defects but also oxygen dopants in monolayer MoS2 are visualized. The method can be readily extended to other two-dimensional systems, as the training is based on unit-cell-level images. Therefore, our results outline novel ways to train the model with minimal data sets, offering great opportunities to fully exploit the power of DL in the materials science community.

Video Unsupervised Domain Adaptation with Deep Learning: A Comprehensive Survey

Video analysis tasks such as action recognition have received increasing research interest with growing applications in fields such as smart healthcare, thanks to the introduction of large-scale datasets and deep learning-based representations. However, video models trained on existing datasets suffer from significant performance degradation when deployed directly to real-world applications due to domain shifts between the training public video datasets (source video domains) and real-world videos (target video domains). Further, with the high cost of video annotation, it is more practical to use unlabeled videos for training. To tackle performance degradation and address concerns in high video annotation cost uniformly, the video unsupervised domain adaptation (VUDA) is introduced to adapt video models from the labeled source domain to the unlabeled target domain by alleviating video domain shift, improving the generalizability and portability of video models. This paper surveys recent progress in VUDA with deep learning. We begin with the motivation of VUDA, followed by its definition, and recent progress of methods for both closed-set VUDA and VUDA under different scenarios, and current benchmark datasets for VUDA research. Eventually, future directions are provided to promote further VUDA research. The repository of this survey is provided at https://github.com/xuyu0010/awesome-video-domain-adaptation.

Self-supervised multi-object tracking based on metric learning

The current paradigm of joint detection and tracking still requires a large amount of instance-level trajectory annotation, which incurs high annotation costs. Moreover, treating embedding training as a classification problem would lead to difficulties in model fitting. In this paper, we propose a new self-supervised multi-object tracking based on the real-time joint detection and embedding (JDE) framework, which we termed as self-supervised multi-object tracking (SS-MOT). In SS-MOT, the short-term temporal correlations between objects within and across adjacent video frames are both considered as self-supervised constraints, where the distances between different objects are enlarged while the distances between same object of adjacent frames are brought closer. In addition, short trajectories are formed by matching pairs of adjacent frames using a matching algorithm, and these matched pairs are treated as positive samples. The distances between positive samples are then minimized for futher the feature representation of the same object. Therefore, our method can be trained on videos without instance-level annotations. We apply our approach to state-of-the-art JDE models, such as FairMOT, Cstrack, and SiamMOT, and achieve comparable results to these supevised methods on the widely used MOT17 and MOT20 challenges.

Wildfire Susceptibility Prediction Based on a CA-Based CCNN with Active Learning Optimization

Wildfires cause great losses to the ecological environment, economy, and people’s safety and belongings. As a result, it is crucial to establish wildfire susceptibility models and delineate fire risk levels. It has been proven that the use of remote sensing data, such as meteorological and topographical data, can effectively predict and evaluate wildfire susceptibility. Accordingly, this paper converts meteorological and topographical data into fire-influencing factor raster maps for wildfire susceptibility prediction. The continuous convolutional neural network (CCNN for short) based on coordinate attention (CA for short) can aggregate different location information into channels of the network so as to enhance the feature expression ability; moreover, for different patches with different resolutions, the improved CCNN model does not need to change the structural parameters of the network, which improves the flexibility of the network application in different forest areas. In order to reduce the annotation of training samples, we adopt an active learning method to learn positive features by selecting high-confidence samples, which contributes to enhancing the discriminative ability of the network. We use fire probabilities output from the model to evaluate fire risk levels and generate the fire susceptibility map. Taking Chongqing Municipality in China as an example, the experimental results show that the CA-based CCNN model has a better classification performance; the accuracy reaches 91.7%, and AUC reaches 0.9487, which is 5.1% and 2.09% higher than the optimal comparative method, respectively. Furthermore, if an accuracy of about 86% is desired, our method only requires 50% of labeled samples and thus saves about 20% and 40% of the labeling efforts compared to the other two methods, respectively. Ultimately, the proposed model achieves the balance of high prediction accuracy and low annotation cost and is more helpful in classifying fire high warning zones and fire-free zones.

Research on semantic segmentation algorithm of high latitude urban river ice based on deep transfer learning

ABSTRACT Automated observation methods for monitoring river ice in high-latitude urban areas are crucial for resource utilization, risk assessment, and navigation. However, current research lacks actual-scale river ice classification, such as low-altitude surveys. This study established a dataset of river ice in the Songhua River near Harbin, Northeast China, using UAV aerial photography and applied the RININet semantic segmentation algorithm for precise classification of different ice types in low-altitude aerial remote sensing images. To address environmental challenges, a feature extraction method integrating channel and spatial attention mechanisms was adopted, along with an improved pyramid pool structure to enhance feature recognition. Additionally, a two-stage transfer learning method established an ice recognition database, overcoming issues like small data volume and high annotation costs. Comparative evaluation metrics demonstrated the high accuracy of the semantic segmentation framework. Furthermore, a method for estimating ice blockage risk was proposed, applicable to various urban river ice management tasks, with practical significance.

WSPolyp-SAM: Weakly Supervised and Self-Guided Fine-Tuning of SAM for Colonoscopy Polyp Segmentation

Ensuring precise segmentation of colorectal polyps holds critical importance in the early diagnosis and treatment of colorectal cancer. Nevertheless, existing deep learning-based segmentation methods are fully supervised, requiring extensive, precise, manual pixel-level annotation data, which leads to high annotation costs. Additionally, it remains challenging to train large-scale segmentation models when confronted with limited colonoscopy data. To address these issues, we introduce the general segmentation foundation model—the Segment Anything Model (SAM)—into the field of medical image segmentation. Fine-tuning the foundation model is an effective approach to tackle sample scarcity. However, current SAM fine-tuning techniques still rely on precise annotations. To overcome this limitation, we propose WSPolyp-SAM, a novel weakly supervised approach for colonoscopy polyp segmentation. WSPolyp-SAM utilizes weak annotations to guide SAM in generating segmentation masks, which are then treated as pseudo-labels to guide the fine-tuning of SAM, thereby reducing the dependence on precise annotation data. To improve the reliability and accuracy of pseudo-labels, we have designed a series of enhancement strategies to improve the quality of pseudo-labels and mitigate the negative impact of low-quality pseudo-labels. Experimental results on five medical image datasets demonstrate that WSPolyp-SAM outperforms current fully supervised mainstream polyp segmentation networks on the Kvasir-SEG, ColonDB, CVC-300, and ETIS datasets. Furthermore, by using different amounts of training data in weakly supervised and fully supervised experiments, it is found that weakly supervised fine-tuning can save 70% to 73% of annotation time costs compared to fully supervised fine-tuning. This study provides a new perspective on the combination of weakly supervised learning and SAM models, significantly reducing annotation time and offering insights for further development in the field of colonoscopy polyp segmentation.

Transfer Learning-Based Specific Emitter Identification for ADS-B over Satellite System

In future aviation surveillance, the demand for higher real-time updates for global flights can be met by deploying automatic dependent surveillance–broadcast (ADS-B) receivers on low Earth orbit satellites, capitalizing on their global coverage and terrain-independent capabilities for seamless monitoring. Specific emitter identification (SEI) leverages the distinctive features of ADS-B data. High data collection and annotation costs, along with limited dataset size, can lead to overfitting during training and low model recognition accuracy. Transfer learning, which does not require source and target domain data to share the same distribution, significantly reduces the sensitivity of traditional models to data volume and distribution. It can also address issues related to the incompleteness and inadequacy of communication emitter datasets. This paper proposes a distributed sensor system based on transfer learning to address the specific emitter identification. Firstly, signal fingerprint features are extracted using a bispectrum transform (BST) to train a convolutional neural network (CNN) preliminarily. Decision fusion is employed to tackle the challenges of the distributed system. Subsequently, a transfer learning strategy is employed, incorporating frozen model parameters, maximum mean discrepancy (MMD), and classification error measures to reduce the disparity between the target and source domains. A hyperbolic space module is introduced before the output layer to enhance the expressive capacity and data information extraction. After iterative training, the transfer learning model is obtained. Simulation results confirm that this method enhances model generalization, addresses the issue of slow convergence, and leads to improved training accuracy.

Review of Semi-Supervised Medical Image Segmentation based on the U-Net

Medical image segmentation is a core task in the field of medical image processing. It aims to separate areas of interest such as organs, tissues, and lesions from the background in medical images to facilitate further analysis, diagnosis, and treatment planning. Accurate image segmentation is crucial for improving the accuracy of disease diagnosis, assessing disease progression, and developing personalized treatment plans. However, fully supervised segmentation methods face the challenge of high annotation costs. With the emergence of the U-Net architecture, semi-supervised medical image segmentation based on U-Net is receiving increasing attention. This article reviews semi-supervised segmentation methods, the design concept and structure of the U-Net network, how it has been extended, and its application in semi-supervised medical image segmentation. The article also identifies the challenges faced by semi-supervised medical image segmentation techniques based on U-Net and speculates on possible future research directions. In conclusion, the article summarizes the potential of semi-supervised medical image segmentation technology based on U-Net as an accurate and efficient tool for medical diagnosis and treatment.

Few-shot biomedical relation extraction using data augmentation and domain information

Relation extraction (RE) plays a pivotal role in biomedical information extraction. However, traditional approaches are often limited by high data annotation costs and extensive time investments. To address this challenge, this study proposes an innovative few-shot learning approach for biomedical RE tasks that utilizes data augmentation and domain information. Specifically, this method enhances the diversity of training data through a synthesized data augmentation strategy. At the same time, it combines domain information with prompt learning techniques. By incorporating domain information, the model’s generalization capability for rare data instances is improved. Furthermore, the prompt learning approach, by integrating prompt templates into the model input, guides the pre-trained language model to more accurately adapt to the RE task. Our model demonstrated exceptional performance in three different biomedical RE tasks, particularly on the I2B2-2010 RE dataset. In the 1-shot, 10-shot, and 50-shot settings, the F1 scores increased by 14.38%, 9.50%, and 6.05%, respectively.

Highly Efficient and Unsupervised Framework for Moving Object Detection in Satellite Videos.

Moving object detection in satellite videos (SVMOD) is a challenging task due to the extremely dim and small target characteristics. Current learning-based methods extract spatio-temporal information from multi-frame dense representation with labor-intensive manual labels to tackle SVMOD, which needs high annotation costs and contains tremendous computational redundancy due to the severe imbalance between foreground and background regions. In this paper, we propose a highly efficient unsupervised framework for SVMOD. Specifically, we propose a generic unsupervised framework for SVMOD, in which pseudo labels generated by a traditional method can evolve with the training process to promote detection performance. Furthermore, we propose a highly efficient and effective sparse convolutional anchor-free detection network by sampling the dense multi-frame image form into a sparse spatio-temporal point cloud representation and skipping the redundant computation on background regions. Coping these two designs, we can achieve both high efficiency (label and computation efficiency) and effectiveness. Extensive experiments demonstrate that our method can not only process 98.8 frames per second on 1024 ×1024 images but also achieve state-of-the-art performance.

MedOptNet: Meta-Learning Framework for Few-Shot Medical Image Classification.

In the medical research domain, limited data and high annotation costs have made efficient classification under few-shot conditions a popular research area. This paper proposes a meta-learning framework, termed MedOptNet, for few-shot medical image classification. The framework enables the use of various high-performance convex optimization models as classifiers, such as multi-class kernel support vector machines, ridge regression, and other models. End-to-end training is then implemented using dual problems and differentiation in the paper. Additionally, various regularization techniques are employed to enhance the model's generalization capabilities. Experiments on the BreakHis, ISIC2018, and Pap smear medical few-shot datasets demonstrate that the MedOptNet framework outperforms benchmark models. Moreover, the model training time is also compared to prove its effectiveness in the paper, and an ablation study is conducted to validate the effectiveness of each module.

Category re-balancing based domain adaptive semantic segmentation

The semantic segmentation task faces the bottleneck of high manual annotation costs. Domain adaptive learning provides an effective solution through inter domain knowledge transfer. However, existing domain adaptive semantic segmentation tasks face the situation of category asymmetry between scenes when dealing with complex scenes. Existing methods lack handling of class imbalance in the prediction process. In domain adaptive semantic segmentation tasks, class imbalance limits the performance of inter domain transfer. Our work focuses on the impact of category distribution on domain adaptive semantic segmentation tasks in image pairs. Inspired by long tail learning, we divided categories into head, middle, and tail categories, and designed effective strategies to enhance the contribution of tail categories in domain adaptation tasks. We introduce the concept of category rebalancing into domain adaptive image matching, and our method is lightweight and effective, proving its effectiveness on general datasets.

Label-Efficient Point Cloud Semantic Segmentation: A Holistic Active Learning Approach

Deep learning models are the state of the art for semantic segmentation of point clouds, the success of which relies on the availability of large-scale annotated datasets. However, it can be prohibitively costly to prepare such datasets. In this work, we propose a holistic active learning (AL) approach to maximize model performance given limited annotation budgets. We investigate the appropriate sample granularity for active selection under the realistic “click” measurement of annotation cost, and demonstrate that superpoint-based selection allows for most efficient usage of the limited budget, when compared with point-level, polygon-level and instance/shape-level selection. We further propose new objective for AL acquisition function and exploit local consistency constraints to boost the performance of our superpoint-based approach. We evaluate our methods on three benchmark datasets, ShapeNet and PartNet and S3DIS. The results demonstrate that AL is an effective strategy to address the high annotation costs in semantic point cloud segmentation.

Improving detection of asphalt distresses with deep learning-based diffusion model for intelligent road maintenance

Research on road infrastructure structural health monitoring is critical due to the increasing problem of deteriorated conditions. The traditional approach to pavement distress detection relies on human-based visual recognition, a time-consuming and labor-intensive method. While Deep Learning-based computer vision systems are the most promising approach, they face the challenges of reduced performance due to the scarcity of labeled data due, high annotation costs misaligned with engineering applications, and limited instances of minority defects. This paper introduces a novel generative diffusion model for data augmentation, creating synthetic images of rare defects. It also investigates methods to enhance image quality and reduce production time. Compared to Generative Adversarial Networks, the optimal configuration excels in reliability, quality and diversity. After incorporating synthetic images into the training of our pavement distress detector, YOLOv5, its mean average precision has been enhanced. This computer-aided system enhances recognition and labelling efficiency, promoting intelligent maintenance and repairs.

Few-shot defect recognition for the multi-domain industry via attention embedding and fine-grained feature enhancement

Defect recognition is an effective quality control measure for the key manufacturing process nodes of industrial products. However, current defect recognition models, which rely on a large amount of supervised data, cannot quickly adapt and identify new classes of defects that appear in the production process. Moreover, automated industrial defect recognition is plagued by the problems of high expert annotation costs, scarce samples, and a wide variety of defects. To address these issues, this study explored a few-shot defect recognition method for the multidomain industry using attention embedding and fine-grained feature enhancement. To enhance the feature representation of highly discriminative regions in images, a spatial multidirectional group attention convolutional backbone network (Res12_SMGA) was proposed. A multi-directional feature-enhanced representation of the group channel space was achieved by grouping the channels and generating direction-sensitive feature mappings. In addition, a lightweight feature enhancement branch (LFEB) was designed to select more discriminative pixel-level features from the enhanced high-dimensional features, while aiming to improve the generalization ability for multidomain industrial few-shot defect recognition tasks. Subsequently, the different granularity feature outputs of Res12_SMGA and LFEB were fused, and defect recognition was completed using the Nearest Class Mean classifier (NCM). Experiments showed that the proposed method achieved optimal performance in identifying 10 types of defects in five industrial product domains. For the 10way-5shot and 5way-5shot settings, the classification accuracies were 91.71% and 95.97%, respectively. Thus, this study demonstrated that new industrial defects can be recognized using only a few labelled samples.

MUE-CoT: multi-scale uncertainty entropy-aware co-training framework for left atrial segmentation

Objective. Accurate left atrial segmentation is the basis of the recognition and clinical analysis of atrial fibrillation. Supervised learning has achieved some competitive segmentation results, but the high annotation cost often limits its performance. Semi-supervised learning is implemented from limited labeled data and a large amount of unlabeled data and shows good potential in solving practical medical problems. Approach. In this study, we proposed a collaborative training framework for multi-scale uncertain entropy perception (MUE-CoT) and achieved efficient left atrial segmentation from a small amount of labeled data. Based on the pyramid feature network, learning is implemented from unlabeled data by minimizing the pyramid prediction difference. In addition, novel loss constraints are proposed for co-training in the study. The diversity loss is defined as a soft constraint so as to accelerate the convergence and a novel multi-scale uncertainty entropy calculation method and a consistency regularization term are proposed to measure the consistency between prediction results. The quality of pseudo-labels cannot be guaranteed in the pre-training period, so a confidence-dependent empirical Gaussian function is proposed to weight the pseudo-supervised loss. Main results. The experimental results of a publicly available dataset and an in-house clinical dataset proved that our method outperformed existing semi-supervised methods. For the two datasets with a labeled ratio of 5%, the Dice similarity coefficient scores were 84.94% ± 4.31 and 81.24% ± 2.4, the HD95 values were 4.63 mm ± 2.13 and 3.94 mm ± 2.72, and the Jaccard similarity coefficient scores were 74.00% ± 6.20 and 68.49% ± 3.39, respectively. Significance. The proposed model effectively addresses the challenges of limited data samples and high costs associated with manual annotation in the medical field, leading to enhanced segmentation accuracy.

Multi-branch Segmentation-guided Attention Network for crowd counting

Crowd counting has gained increasing attention recently owing to its importance in public safety. However, it remains a challenging task due to background complexities and high annotation costs. To address these challenges, we propose the Multi-branch Segmentation-guided Attention Network (MSGANet). To deal with the complex background, we incorporate segmentation-guided attention branches into both shallow and deep layers of the baseline model, allowing simultaneous consideration of spatial and semantic information. Multi-level attention maps enable the network to minimize the influence of complex backgrounds while focusing on the regions containing crowds. To tackle the issue of costly annotations, MSGANet utilizes only point annotations to generate ground truth density and segmentation maps, eliminating additional expenses. Our results demonstrate that our approach achieves highly competitive performance compared to state-of-the-art crowd counting methods.

Early Prediction of Covid-19 Samples from Chest X-ray Images using Deep Learning Approach

Aims: In this study, chest X-ray (CXR) and computed tomography (CT) images are used to analyse and detect COVID-19 using an unsupervised deep learning-based feature fusion approach. Background: The reverse transcription-polymerase chain reaction (RT-PCR) test, which has a reduced viral load, sampling error, etc., is used to detect COVID-19, which has sickened millions of people worldwide. It is possible to check chest X-rays and computed tomography scans because the majority of infected persons have lung infections. The COVID-19 diagnosis can be made early using both CT and CXR imaging modalities, which is an alternative to the RT-PCR test. Objective: The manual diagnosis of CXR pictures and CT scans is labor and time-intensive. Many AI-based solutions are being investigated to tackle this problem, including deep learning-based detection models, which can be utilized to assist the radiologist in making a more accurate diagnosis. However, because of the demand for specialized knowledge and high annotation costs, the amount of annotated data available for COVID-19 identification is constrained. Additionally, the majority of current cutting-edge deep learning-based detection models use supervised learning techniques. Because a tagged dataset is not required, we have investigated various unsupervised learning models for COVID-19 identification in this work. Methods: In this study, we suggest a COVID-19 detection method based on unsupervised deep learning that makes use of the feature fusion technique to improve performance. Based on this an automated CNN model is built for the detection of COVID-19 samples from healthy and pneumonic cases using chest X-ray images. Results: This model has scored an accuracy of about 99% for the classification between covid positive and covid negative. Based on this result further classification will be done for pneumonic and non-pneumonic which has scored an accuracy of 94%. Conclusion: On both datasets, the COVID-19 detection method based on feature fusion and deep unsupervised learning showed promising results. Additionally, it outperforms four well-known unsupervised methods already in use.

Research on the construction of event corpus with document-level causal relations for social security

Event corpora are imperative to train event extraction models. Currently, most existing event corpora suffer from being available only in English, and their construction is limited by high annotation costs. This paper aims to construct a corpus that concerns social security causality events in Chinese and proposes a faster and less expensive construction method. The contributions are as follows: (i) An event corpus SSECau for the social security field in Chinese is constructed. They are from 2,235 web texts and microblogs, with event causality annotated at the document level. (ii) A corpus construction method with manual annotation and machine pre-tagging is proposed to improve accuracy and speed. (iii) A pre-tagging method based on BiLSTM-CRF (bidirectional long short-term memory and conditional random field) is deployed to extract events automatically. The experimental results show the best consistency between automatic pre-tagging and manual annotation can reach up to 82 %; while the dynamic tagging process improves both the labeling speed and accuracy. The SSECau corpus can aid the development and evaluation of event extraction models for the social security field; annotated cause-effect relationships at the document level can potentially enhance the training of complex extraction models; the proposed dynamic process with pre-tagging can serve as a reference for future corpus construction.