Automatic Labeling System Research Articles

CORAL: Collaborative Automatic Labeling System Based on Large Language Models

In the era of big data, data annotation is integral to numerous applications. However, it is widely acknowledged as a laborious and time-consuming process, significantly impeding the scalability and efficiency of data-driven applications. To reduce the human cost, we demonstrate CORAL, a collaborative automatic labeling system driven by large language models (LLMs), which achieves high-quality annotation with the least human effort. Firstly, CORAL employs LLM to automatically annotate vast datasets, generating coarse-grained labels. Subsequently, a weakly-supervised learning module trains small language models (SLMs) using noisy label learning techniques to distill accurate labels from LLM's annotations. It also allows statistical analysis of model outcomes to identify potentially erroneous labels, reducing the human cost of error detection. Furthermore, CORAL supports iterative refinement by LLMs and SLMs using manually corrected labels, thereby ensuring continual enhancement in annotation quality and model performance. A visual interface enables annotation process monitoring and result analysis.

AutoDLAR: A Semi-supervised Cross-modal Contact-free Human Activity Recognition System

WiFi-based human activity recognition (HAR) plays an essential role in various applications such as security surveillance, health monitoring, and smart home. Existing HAR methods, though yielding promising performance in indoor scenarios, highly depend on a massive labeled dataset for training which is extremely difficult to acquire in practical applications. In this paper, we present an automatic data labeling and HAR system, termed AutoDLAR. Taking a semi-supervised cross-modal learning framework with a hybrid loss function as the core, AutoDLAR transfers rich visual information to automatically label WiFi signals for WiFi-based HAR. Specifically, we devise a lightweight and multi-view WiFi sensing model with a parallel feature embedding method to accurately identify activities and accelerate recognition speed. Then, we exploit the video data to fine-tune a well-established visual HAR model, generating effective pseudo-labels for guiding the WiFi model’s training. We also build a synchronized Video-WiFi dataset with seven types of human activities under different scenarios to enable training and validating the semi-supervised HAR system. Extensive experiments on our collected activity dataset and the emotion recognition benchmark demonstrate that AutoDLAR attains an average accuracy of over 95.89% without manual labeling and only spends the inference time of 3.35 ms, outperforming the state-of-the-art (SOTA) methods.

Enhancing Fish Auction with Deep Learning and Computer Vision: Automated Caliber and Species Classification

The accurate labeling of species and size of specimens plays a pivotal role in fish auctions conducted at fishing ports. These labels, among other relevant information, serve as determinants of the objectivity of the auction preparation process, underscoring the indispensable nature of a reliable labeling system. Historically, this task has relied on manual processes, rendering it vulnerable to subjective interpretations by the involved personnel, therefore compromising the value of the merchandise. Consequently, the digitization and implementation of an automated labeling system are proposed as a viable solution to this ongoing challenge. This study presents an automatic system for labeling species and size, leveraging pre-trained convolutional neural networks. Specifically, the performance of VGG16, EfficientNetV2L, Xception, and ResNet152V2 networks is thoroughly examined, incorporating data augmentation techniques and fine-tuning strategies. The experimental findings demonstrate that for species classification, the EfficientNetV2L network excels as the most proficient model, achieving an average F-Score of 0.932 in its automatic mode and an average F-Score of 0.976 in its semi-automatic mode. Concerning size classification, a semi-automatic model is introduced, where the Xception network emerges as the superior model, achieving an average F-Score of 0.949.

Dynamic Automated Labeling System for Real-Time User Intention Analysis

Dynamic Automated Labeling System for Real-Time User Intention Analysis

A Cluster-then-label Approach for Few-shot Learning with Application to Automatic Image Data Labeling

Few-shot learning (FSL) aims at learning to generalize from only a small number of labeled examples for a given target task. Most current state-of-the-art FSL methods typically have two limitations. First, they usually require access to a source dataset (in a similar domain) with abundant labeled examples, which may not always be possible due to privacy concerns and copyright issues. Second, they typically do not offer any estimation of the generalization error on the target FSL task, because the handful of labeled examples must be used for training and cannot spare a validation subset. In this article, we propose a cluster-then-label approach to perform few-shot learning. Our approach does not require access to the labeled source dataset and provides an estimation of generalization error. We show empirically, on four benchmark datasets, that our approach provides competitive predictive performance to state-of-the-art FSL approaches and our generalization error estimation is accurate. Finally, we explore the application of our proposed method to automatic image data labeling. We compare our method with existing automatic data labeling systems. The end-to-end performance of our method outperforms the state-of-the-art automatic data labeling system Snuba by 26% and is only 7% away from the fully supervised upper bound.

Large-scale real-world radio signal recognition with deep learning

In the past ten years, many high-quality datasets have been released to support the rapid development of deep learning in the fields of computer vision, voice, and natural language processing. Nowadays, deep learning has become a key research component of the Sixth-Generation wireless systems (6G) with numerous regulatory and defense applications. In order to facilitate the application of deep learning in radio signal recognition, in this work, a large-scale real-world radio signal dataset is created based on a special aeronautical monitoring system - Automatic Dependent Surveillance-Broadcast (ADS-B). This paper makes two main contributions. First, an automatic data collection and labeling system is designed to capture over-the-air ADS-B signals in the open and real-world scenario without human participation. Through data cleaning and sorting, a high-quality dataset of ADS-B signals is created for radio signal recognition. Second, we conduct an in-depth study on the performance of deep learning models using the new dataset, as well as comparison with a recognition benchmark using machine learning and deep learning methods. Finally, we conclude this paper with a discussion of open problems in this area.

A Two-Phase Fashion Apparel Detection Method Based on YOLOv4

Object detection is one of the important technologies in the field of computer vision. In the area of fashion apparel, object detection technology has various applications, such as apparel recognition, apparel detection, fashion recommendation, and online search. The recognition task is difficult for a computer because fashion apparel images have different characteristics of clothing appearance and material. Currently, fast and accurate object detection is the most important goal in this field. In this study, we proposed a two-phase fashion apparel detection method named YOLOv4-TPD (YOLOv4 Two-Phase Detection), based on the YOLOv4 algorithm, to address this challenge. The target categories for model detection were divided into the jacket, top, pants, skirt, and bag. According to the definition of inductive transfer learning, the purpose was to transfer the knowledge from the source domain to the target domain that could improve the effect of tasks in the target domain. Therefore, we used the two-phase training method to implement the transfer learning. Finally, the experimental results showed that the mAP of our model was better than the original YOLOv4 model through the two-phase transfer learning. The proposed model has multiple potential applications, such as an automatic labeling system, style retrieval, and similarity detection.

Integration of the Codonics Safe Label System® and the Omnicell XT® Anesthesia Workstation into Pediatric Anesthesia Practice: Utilizing Technology to Increase Medication Labeling Compliance and Decrease Medication Discrepancies While Maintaining User Acceptability.

Background: Perioperative medication errors are recognized as a source of patient morbidity and mortality. Medication management systems with built-in scanning and label-printing functions that integrate with medication-dispensing cabinets have the potential to decrease medication administration errors by improving compliance with medication labeling. Whether these management systems will also improve periodic automatic replacement (PAR) inventory control and be accepted by users is unknown. We hypothesized that implementation of the Codonics Safe Label System®, an automated labeling system (ALS), would increase compliance with labeling guidelines and improve PAR inventory control by decreasing medication discrepancies while maintaining user acceptability in the OR. Methods: We audited a cohort of anesthesia workstations and electronic anesthesia records for 2 months to compare dispensed and administered medications and establish a discrepancy baseline. We also observed a convenience sample of syringes to evaluate labeling compliance. Post-implementation of the ALS, we repeated the audit. Finally, an anonymous survey was distributed electronically to providers to assess user acceptability. Results: Pre-implementation the average daily medication discrepancy rate was 9.7%, decreasing to 6.1% post-implementation (χ2 1 = 43.9; P < .0001). Pre-implementation 330 of 696 syringes (47.4%) were either missing a label or labeling elements. After implementation, 100% of all syringes received a label with the complete required labeling information (P < .0001). All respondents agreed or strongly agreed that the system was easy to use, accurate, met their needs, printed labels quickly, improved safety and efficiency, and was recommendable. Conclusion: The ALS significantly increased the rate of best-practice-compliant medication labeling while reducing medication inventory discrepancies. The system was highly accepted by providers.

Domain Independent Automatic Labeling system for Large-scale Social Data using Lexicon and Web-based Augmentation

Recently, with the large-scale adoption of social media, people have begun to express their opinion on these sites in the form of reviews. Potential consumers often forced to wade through huge amount of reviews to make informed decision. Sentiment analysis has become rapid and effective way to automatically gauge consumers’ opinion. However, such analysis often requires tedious process of manual tagging of large training examples or manually building a lexicon for the purpose of classifying reviews as positive or negative. In this paper, we present a method to automate the tedious process of labeling large textual data in an unsupervised, domain independent and scalable manner. The proposed method combines the lexicon-based and Web-based Point Wise Mutual Information (PMI) statistics to find the Semantic Orientation (SO) of opinion expressed in a review. Based on proposed methods a system called Domain Independent Automatic Labeling System (DIALS) has been implemented, which takes collection of text from any domain as input and generates fully labeled dataset in an unsupervised and scalable manner. The result generated can be used to track and summarize online discussion and/or use to train any classifier in the next stage of development. The effectiveness of system is tested by comparing it with baseline machine learning and lexicon-based methods. Experiments on multi-domains dataset has shown that proposed method consistently shown improved recall and accuracy as compared to baseline machine learning and lexicon-based methods.

Abnormal Alterations of Cortical Thickness in 16 Patients with Type 2 Diabetes Mellitus: A Pilot MRI Study.

Objective The aim of this study is to investigate the cerebral cortical thickness changes in type 2 diabetes mellitus (T2DM) using a whole brain cortical thickness mapping system based on brain magnetic resonance imaging (MRI).Methods High resolution three-dimensional T1-weighted fast spoiled gradient recalled echo MR images were obtained from 16 patients with T2DM, as well as from 16 normal controls. The whole brain cortical thickness maps were generated, and the cortical thickness of each brain region was calculated according to gyral based regions of interest (ROI) using an automated labeling system by the Freesurfer software. We compared mean cortical thickness at each brain region by the analysis of covariance with age and sex as covariates. The regional difference of the cortical thickness over the whole brain was compared by the analysis of surface-based cortical thickness.Results Mean cortical thicknesses analysis showed bilateral cerebrum in the patients with T2DM (left: 2.52±0.07 mm; right: 2.51±0.08 mm) were significant thinner than those in the normal controls (left: 2.56±0.09 mm; right: 2.56±0.09 mm) (both P<0.05). Regional cortical thinning in T2DM was demonstrated in the paracentral lobule, postcentral gyrus, lateral occipital gyrus, lingual gyrus, precuneus, superior temporal gyrus, middle temporal gyrus, inferior temporal gyrus and posterior cingulate gyrus, compared to the normal controls. The cortical thickness of left middle cingulate and right inferior temporal gyrus were negatively correlated with the disease course.Conclusion A widespread cortical thinning was revealed in patients with T2DM by the analysis of brain cortical thickness on MR. Our finding supports the idea that T2DM could lead to subtle diabetic brain structural changes.

Improvement of Preventive Maintenance Plan for an Automatic Labeling System: A Case Study of a Carbonated Soft Drink Industry

The aims of this research are to improve the preventive maintenance plan and to prepare the autonomous maintenance plan of an automatic labeling system for a carbonated soft drink industry as a case study. The labeling system of the case study has experienced the wastes in maintenance. According to the maintenance problem analysis, the main causes are lack of or inadequate preventive maintenance program, inadequate work order monitoring system, and inadequate staff skills (problem-solving skills of maintenance) to perform work. Therefore, in order to reduce the number and length of machine breakdowns and the seriousness of breakdowns, the preventive maintenance program is used instead of breakdown and routine maintenance by which maintenance skills are required, and in order to improve how machines are operated and handled by improving the skills, and the autonomous maintenance manual is set up to prepare for autonomous maintenance. After the improved preventive maintenance plan has been developed and then implemented, it was found that machine downtime reduced by 1.35% and machine availability increased by 1.74%.

Precrash Classification of Car Accidents for Improved Occupant Safety Systems

Abstract New occupant safety systems, which adapt their behavior to the severity of an accident, may improve vehicle safety. We present an approach to learn a prediction model, which estimates crash severity prior to collision. Based on accident parameters acquired with precrash sensors, the learned model categorizes impending accidents into one of multiple severity classes. Besides describing an automatic labeling system for data preparation, we investigate the performance of different classifiers. Results on simulation data demonstrate a classification performance of 84% correctly classified test cases. We discuss a potential application and finish with ideas for classification improvement.

Sistema de etiquetado automático de contenidos educativos para favorecer su reusabilidad e interoperabilidad

En este trabajo presentamos un proyecto de etiquetaje automático de objetos de aprendizaje. Destacando la importancia del etiquetado de materiales para favorecer su reusabilidad, se realiza un análisis de los problemas que nos encontramos para añadir metadatos a los recursos digitales. También se hace un estudio de alternativas al etiquetado manual de recursos así como las ventajas de un buen etiquetado de materiales. Como aplicación práctica se hace una aproximación a lo que sería un sistema de etiquetado automático de materiales, dando pautas sobre cómo montar un sistema autónomo de recopilación, etiquetado y organización de objetos de aprendizaje.

Automated identification of thoracolumbar vertebrae using orthogonal matching pursuit

A computer-assisted system that can automatically provide rapid localization and accurate labeling of vertebral disks and bodies is a highly desirable tool due to the large demand for the diagnostic imaging and surgical planning of the vertebral column structures. However, a reliable detection and definitive labeling of vertebrae can be difficult due to factors such as the limited imaging coverage and various vertebral anomalies particularly in the thoracolumbar and lumbosacral junctions. In this paper, we investigate the problem of identifying the last thoracic and first lumbar vertebrae in CT images. The main purpose of this study is to improve the accuracy of labeling vertebrae of an automatic spine labeling system especially when the field of view is limited in the lower spine region. We present a dictionary-based classification method using a cascade of simultaneous orthogonal matching pursuit classifiers on 2D vertebral regions extracted from the maximum intensity projection images. The performance of the proposed method in terms of accuracy and speed has been validated by experimental results on hundreds of CT images collected from various clinical sites.

Changes in the Frontotemporal Cortex and Cognitive Correlates in First-Episode Psychosis

BackgroundLoss of cortical volume in frontotemporal regions has been reported in patients with schizophrenia and their relatives. Cortical area and thickness are determined by different genetic processes, and measuring these parameters separately may clarify disturbances in corticogenesis relevant to schizophrenia. Our study also explored clinical and cognitive correlates of these parameters.MethodsThirty-seven patients with first-episode psychosis (34 schizophrenia, 3 schizoaffective disorder) and 38 healthy control subjects matched for age and sex took part in the study. Imaging was performed on an magnetic resonance imaging 1.5-T scanner. Area and thickness of the frontotemporal cortex were measured using a surface-based morphometry method (Freesurfer). All subjects underwent neuropsychologic testing that included measures of premorbid and current IQ, working and verbal memory, and executive function.ResultsReductions in cortical area, more marked in the temporal cortex, were present in patients. Overall frontotemporal cortical thickness did not differ between groups, although regional thinning of the right superior temporal region was observed in patients. There was a significant association of both premorbid IQ and IQ at disease onset with area, but not thickness, of the frontotemporal cortex, and working memory span was associated with area of the frontal cortex. These associations remained significant when only patients with schizophrenia were considered.ConclusionsOur results suggest an early disruption of corticogenesis in schizophrenia, although the effect of subsequent environmental factors cannot be excluded. In addition, cortical abnormalities are subject to regional variations and differ from those present in neurodegenerative diseases.

Subjective Loudness and Reality of Auditory Verbal Hallucinations and Activation of the Inner Speech Processing Network

One of the most influential cognitive models of auditory verbal hallucinations (AVH) suggests that a failure to adequately monitor the production of one's own inner speech leads to verbal thought being misidentified as an alien voice. However, it is unclear whether this theory can explain the phenomenological complexity of AVH. We aimed to assess whether subjective perceptual and experiential characteristics may be linked to neural activation in the inner speech processing network. Twenty-two patients with schizophrenia and AVH underwent a 3-T functional magnetic resonance imaging scan, while performing a metrical stress evaluation task, which has been shown to activate both inner speech production and perception regions. Regions of interest (ROIs) comprising the putative inner speech network were defined using the Anatomical Automatic Labeling system. Correlations were calculated between scores on the "loudness" and "reality" subscales of the Auditory Hallucination Rating Scale (AHRS) and activation in these ROIs. Second, the AHRS subscales, and general AVH severity, indexed by the Positive and Negative Syndrome Scale, were correlated with a language lateralization index. Louder AVH were associated with reduced task-related activity in bilateral angular gyrus, anterior cingulate gyrus, left inferior frontal gyrus, left insula, and left temporal cortex. This could potentially be due to a competition for shared neural resources. Reality on the other hand was found to be associated with reduced language lateralization. Strong activation of the inner speech processing network may contribute to the subjective loudness of AVH. However, a relatively increased contribution from right hemisphere language areas may be responsible for the more complex experiential characteristics, such as the nonself source or how real AVH are.

Gestures Orchestrate Brain Networks for Language Understanding

Although the linguistic structure of speech provides valuable communicative information, nonverbal behaviors can offer additional, often disambiguating cues. In particular, being able to see the face and hand movements of a speaker facilitates language comprehension [1]. But how does the brain derive meaningful information from these movements? Mouth movements provide information about phonological aspects of speech [2-3]. In contrast, cospeech gestures display semantic information relevant to the intended message [4-6]. We show that when language comprehension is accompanied by observable face movements, there is strong functional connectivity between areas of cortex involved in motor planning and production and posterior areas thought to mediate phonological aspects of speech perception. In contrast, language comprehension accompanied by cospeech gestures is associated with tuning of and strong functional connectivity between motor planning and production areas and anterior areas thought to mediate semantic aspects of language comprehension. These areas are not tuned to hand and arm movements that are not meaningful. Results suggest that when gestures accompany speech, the motor system works with language comprehension areas to determine the meaning of those gestures. Results also suggest that the cortical networks underlying language comprehension, rather than being fixed, are dynamically organized by the type of contextual information available to listeners during face-to-face communication.

1A1-L09 人の知識や経験を反映したセンサ空間居住者行動ラベリング法(人間機械協調)

We developed an automatic behavior labeling system for tremendous data such as life-logs gathered through sensing human life. In this paper, the following advantages of the system were emphasized: 1) labeling behavior exploiting human knowledge and experience 2) relaxing constraints and selecting features freely as to constructing a labeling model, 3) helping find effective features for each behavior label, 4) dealing with incomplete data efficiently. The sequential labeling model with above features could be realized using Conditional Random Field (CRF) framework. We describe features of CRF, procedures of selecting effective features and how to deal with incomplete data in CRF framework. Lastly the model with CRF framework was applied to the real sensing environment and we state the performance based on the model and the feasibility of CRF framework in the human behavior recognition fields.

An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest

In this study, we have assessed the validity and reliability of an automated labeling system that we have developed for subdividing the human cerebral cortex on magnetic resonance images into gyral based regions of interest (ROIs). Using a dataset of 40 MRI scans we manually identified 34 cortical ROIs in each of the individual hemispheres. This information was then encoded in the form of an atlas that was utilized to automatically label ROIs. To examine the validity, as well as the intra- and inter-rater reliability of the automated system, we used both intraclass correlation coefficients (ICC), and a new method known as mean distance maps, to assess the degree of mismatch between the manual and the automated sets of ROIs. When compared with the manual ROIs, the automated ROIs were highly accurate, with an average ICC of 0.835 across all of the ROIs, and a mean distance error of less than 1 mm. Intra- and inter-rater comparisons yielded little to no difference between the sets of ROIs. These findings suggest that the automated method we have developed for subdividing the human cerebral cortex into standard gyral-based neuroanatomical regions is both anatomically valid and reliable. This method may be useful for both morphometric and functional studies of the cerebral cortex as well as for clinical investigations aimed at tracking the evolution of disease-induced changes over time, including clinical trials in which MRI-based measures are used to examine response to treatment.

음성학적 지식과 DAC 기반 분할 알고리즘

본 논문에서는 음절이 잘 발달되어 있는 한국어에 대해서 신뢰할 수 있는 완전 자동화된 레이블링 시스템을 제안한다. 음운 및 음향학적인 정보를 최대한 이용하고 분할에러를 줄이기 위해서 조절 메카니즘의 하나로 DAC개념을 사용하여 음성을 speechlet으로 나누고 분할 된 음성 구간에 대해서 레이블링을 시도하는 DAC기반 분할알고리즘이다. HMM방법이 획일적이고 확정적인 성능을 갖는 반면 본 제안 방법은 음성학적인 특화지식을 컴포넌트로 개발 추가 계속 향상시킬 수 있는 프레임워크를 제시하고 있다는 점에서 주요 의의가 있다고 하겠다. MM과 같은 통계학적인 방법을 이용하지 않고 음운학적, 음향학적 지식만을 이용하는 새로운 방법은 수행속도와 음성학적인 특화 지식컴포넌트를 확장함에 따라 일관성이 있으며 효과적 방법으로 적용가능 할 것이다. 제안 방법을 검증하기 위하여 실험결과를 제시하였다. This paper presents a reliable fully automatic labeling system which fits well with languages having well-developed syllables such as in Korean. The ASL System utilize DAC (Divide and Conquer), a control mechanism, based segmentation algorithm to use phonetic and acoustic information with greater efficiency. The segmentation algorithm is to devide speech signals into speechlets which is localized speech signal pieces and to segment each speechlet for speech boundaries. While HMM method has uniform and definite efficiencies, the suggested method gives framework to steadily develope and improve specified acoustic knowledges as a component. Without using statistical method such as HMM, this new method use only phonetic-acoustic information. Therefore, this method has high speed performance, is consistent extending the specific acoustic knowledge component, and can be applied in efficient way. we show experiment result to verify suggested method at the end.