Automatic Acquisition Research Articles

Climbing Mont Blanc (4808 m a.s.l.): What Are the Risks in the Grand Couloir Du Goûter and On the North Face of Mont Blanc Du Tacul?

Mountaineering is a sport highly dependent on the conditions of the physical environments in which it takes place. As a result, mountaineers have to deal with major glacial and periglacial hazards, such as those affecting the two main routes used to climb Mont Blanc (4808 m a.s.l.; Western European Alps): rockfalls in the Grand Couloir du Goûter on the Voie Royale and serac falls on the NNW slope of Mont Blanc du Tacul on the Trois Monts route. Hazards in these two sectors and the number of climbers were measured using a multi-sensor system and automatic photographic acquisition, respectively. On the first site, which has a very high accident rate with an average of 3.7 deaths per year, 17,768 passages were recorded during the summer of 2019 and 2,648 seismic signals were classified as rockfalls affecting the Grand Couloir. The average risk of death associated with these rockfalls, the frequency of which depends on temperatures and the presence of liquid water in the ground, is estimated at 1.7 × 10-4. At Tacul, where an average of 0.6 deaths occur each year, 6,770 passages were recorded during the summer of 2017, as well as 31 serac falls unrelated to temperatures. The risk of death is 8.3 × 10-5. The risk is also quantified for each track made on the glacier. The results obtained makes it possible to specify the risks experienced by climbers and to promote risk reduction strategies.

A high-quality dataset construction method for text mining in materials science

Numerous data and knowledge generated and stored as text in peer-reviewed scientific literature are important for materials research and development. Although text mining can automatically explore this information, the barriers of acquiring high-quality textual data prevent its general application in materials science. Herein, we systematically analyze the issues of textual DATA QUALITY and related research from the perspectives of data quality and quantity. Following this, we propose a pipeline to construct high-quality datasets for text mining in materials science. In this pipeline, we utilize the traceable automatic acquisition scheme of literature to ensure the traceability of textual data. Then, a data processing method driven by downstream tasks is used to generate high-quality pre-annotated corpora conditioned on the characteristics of material texts. On this basis, we define a general annotation scheme derived from materials science tetrahedron to complete high-quality annotation. Finally, a conditional data augmentation model incorporating material domain knowledge (cDA-DK) is constructed to augment the data quantity. Experimental results on datasets with various material systems demonstrate that our method can effectively improve the accuracy of downstream models and the F1-score towards the named entity recognition task in NASICON-type solid electrolyte material reaches 84%. This study provides an important insight into the general application of text mining in materials science, and is expected to advance the material design and discovery driven by data and knowledge bidirectionally.

Research on the correlation between three-dimensional morphology and temperature changes in potato slices during drying

Drying is an effective method to reduce potato storage loss. However, potatoes have high porosity with high water content. Shrinkage during drying can lead to folding and cracking of the dried product form. Therefore, this study explored the correlation between the 3D morphology and temperature distribution changes of potato slices during drying, with the aim of providing a reference for the detection of quality changes. An online automatic acquisition device to obtain 3D morphology and temperature information was designed and built. Hot air-drying experiments were conducted on the potato slices. 3D morphology images and temperature images of the potato slices were acquired by 3D and temperature sensors, and the two images were registered using the random sample consensus (RANSAC) algorithm. The region of interest of each image was extracted by algorithms such as threshold segmentation, hole filling and morphological erosion, and the 3D morphology information and temperature information were obtained. The mapping, range and average of each acquisition point were calculated for correlation analysis. Spearman's rank correlation coefficients and Maximum Information Coefficient (MIC) values were selected as measures for the correlation study. The results showed that the Spearman's rank correlation coefficients between average height and average temperature were mostly above 0.7 in absolute value, and the MICs were mostly above 0.9. The average values of the 3D information and temperature information exhibited an extremely strong correlation. This paper gives a new approach to investigate the morphological changes in the drying process by quantifying the relationship between 3D morphology and temperature distribution. This can guide the improvement of potato drying and processing methods.

Design and implementation of a smart Internet of Things chest pain center based on deep learning.

The data input process for most chest pain centers is not intelligent, requiring a lot of staff to manually input patient information. This leads to problems such as long processing times, high potential for errors, an inability to access patient data in a timely manner and an increasing workload. To address the challenge, an Internet of Things (IoT)-driven chest pain center is designed, which crosses the sensing layer, network layer and application layer. The system enables the construction of intelligent chest pain management through a pre-hospital app, Ultra-Wideband (UWB) positioning, and in-hospital treatment. The pre-hospital app is provided to emergency medical services (EMS) centers, which allows them to record patient information in advance and keep it synchronized with the hospital's database, reducing the time needed for treatment. UWB positioning obtains the patient's hospital information through the zero-dimensional base station and the corresponding calculation engine, and in-hospital treatment involves automatic acquisition of patient information through web and mobile applications. The system also introduces the Bidirectional Long Short-Term Memory (BiLSTM)-Conditional Random Field (CRF)-based algorithm to train electronic medical record information for chest pain patients, extracting the patient's chest pain clinical symptoms. The resulting data are saved in the chest pain patient database and uploaded to the national chest pain center. The system has been used in Liaoning Provincial People's Hospital, and its subsequent assistance to doctors and nurses in collaborative treatment, data feedback and analysis is of great significance.

CLEMSite, a software for automated phenotypic screens using light microscopy and FIB-SEM.

In recent years, Focused Ion Beam Scanning Electron Microscopy (FIB-SEM) has emerged as a flexible method that enables semi-automated volume ultrastructural imaging. We present a toolset for adherent cells that enables tracking and finding cells, previously identified in light microscopy (LM), in the FIB-SEM, along with the automatic acquisition of high-resolution volume datasets. We detect the underlying grid pattern in both modalities (LM and EM), to identify common reference points. A combination of computer vision techniques enables complete automation of the workflow. This includes setting the coincidence point of both ion and electron beams, automated evaluation of the image quality and constantly tracking the sample position with the microscope's field of view reducing or even eliminating operator supervision. We show the ability to target the regions of interest in EM within 5 µm accuracy while iterating between different targets and implementing unattended data acquisition. Our results demonstrate that executing volume acquisition in multiple locations autonomously is possible in EM.

An Alternative Experimental Procedure to Determine the Solubility of Potassium Nitrate in Water with Automatic Data Acquisition Using Arduino for Secondary School: Development and Validation with Pre-Service Chemistry Teachers

In the laboratory, data acquisition systems are important, as they allow us to easily and precisely collect data. In this sense, Arduino emerges as an automatic data acquisition device with great potential, due to its low cost and high versatility. In this work, we describe the development of an experimental apparatus, with automatic data acquisition using Arduino, to determine the variation of the solubility of potassium nitrate in water as a function of temperature. Ten chemistry teachers in initial training were involved in developing and validating that alternative experimental procedure for secondary school. Pre-service chemistry teachers determined the solubility of the aforementioned salt at different temperatures, using both a method for the study of the solubility of salts that does not resort to automatic data acquisition as well as the alternative method proposed in this work. The experimental solubility curves of potassium nitrate were plotted for both situations. The experimental results obtained by both methods are similar and very close to the values reported in the literature. Moreover, chemistry teachers in initial training recognize that the proposed method can promote the development of secondary students’ skills such as greater mastery in assembling electrical circuits and in the use of technological devices or software for automatic data acquisition and processing. Thus, the results suggest the feasibility of the developed experimental method for its implementation in an educational context with secondary students and prove it to be an asset for the education of students, when compared to the traditionally used method.

Geometry-informed deep learning-based structural component segmentation of post-earthquake buildings

Traditional manual visual inspections have demonstrated certain shortcomings in post-earthquake assessment of urban buildings, such as being time-consuming and laborious. In contrast, computer vision (CV) and unmanned aerial vehicle (UAV) approaches have revealed competitive potentials in the fields of automatic data acquisition, data processing, and autonomous decision-making. In UAV images, structural components of post-earthquake buildings often present different scales, which are affected by different local damage. Therefore, acquiring the feature information of structural components has precisely been significant for refined damage assessment of post-earthquake buildings. This study proposes a geometry-informed deep learning-based structural component segmentation of post-earthquake buildings. An Enhanced UNet model is established with a new synthetical loss function containing the geometric consistency (GC) term. Given an edge closure of a connected domain for homogeneous structural components, the GC term comprises split line loss and area loss to adapt to the circumference and area constraints of each component region. The Enhanced UNet network is designed to improve the extraction capability of high-level features, and it includes six encoder stages (superior to five in the original version), of which the bottom four stages have many convolution layers, and five corresponding decoders. The investigated synthetic QuakeCity dataset includes 4,809 images with a resolution of 1,920 × 1,080 pixels. Training and test results reveal that compared to the original UNet, the proposed method achieves a more stable training process and higher test accuracy for structural component segmentation. The proposed method can achieve a mIoU of 97.97 %, which is 1.29 % higher than that of the original UNet. In addition, misrecognition of inner voids inside structural components is addressed, which further validates the optimization efficiency of the proposed geometric constraints. Ablation experiments are conducted to confirm the effectiveness of the proposed GC loss and Enhanced UNet network. The proposed method shows good generalization ability in robustness tests in complex real-world scenarios under various disturbances, including abnormal exposure and rain lines in various intensities.

1237: NOVEL ANALYSIS OF ECHOCARDIOGRAM INDICES AND THEIR ASSOCIATION WITH OUTCOMES IN PEDIATRIC SEPSIS

Introduction: Sepsis induced myocardial dysfunction is common in pediatric sepsis. We sought to utilize novel, specialized algorithms of echocardiogram data that allow for automatic acquisition of echocardiographic indices from all four chambers without advanced interpretation by cardiologists. We hypothesized that these algorithms could predict mortality and morbidity in septic children. Methods: A retrospective cohort study of 59 children admitted to the ICU with sepsis and echocardiogram data. Three user inputted points outlining each heart chamber were applied to existing echocardiograms. Novel algorithms then computed the relevant echocardiographic indices. We examined right and left sided indices for systolic function [ejection fraction (EF), stroke volume (SV) systolic wave (S’)], diastolic function [early diastolic annular velocity (E’), late diastolic annular velocity (A’)], chamber sizes, and strain indices. Our primary outcome was mortality, secondary outcomes included ventilator free days (VFD) at 28 days, and length of stay (LOS). Linear regression models were constructed. Results: Six (10.2%) of 59 patients died. Baseline characteristics were similar between survivors and nonsurvivors. Survivors had higher right ventricular (RV) SV [11.2 mL (4.7, 52.3) versus 3.04 mL (2.1, 4.2), p=0.010], RV (S’) [6.2 cm/s (5.0, 7.6) versus 4.5 cm/s (2.9, 6.2), p=0.046] and RV end-diastolic volume (EDV) [29.6 mL (12.3, 70.9) versus 9.1 mL (5.8, 17.6), p=0.016]. Multiple right chamber indices were associated with more VFD at 28 days [RV S’: β 2.7 (1.3 to 4.1), p=< 0.000, RV E’: β 2.3 (0.9 to 3.6), p=0.002, right atrial (RA) SV: β 0.3 (0.1 to 0.5), p=0.010, RA EF: β 0.2 (0.0 to 0.4) p=0.047, and RA EDV: β 0.4 (0.1 to 0.8) p=0.017]. Multiple diastolic indices were associated with differences in both ICU LOS in hours [left ventricular (LV) A’: β -129.5 (-208.4 to -50.6), p=0.020; RV S’: β -96.1 (-168.7 to -23.4), p=0.01] and hospital LOS in hours [LV A’: β -144.9 (-266.6 to -23.4), p=0.02; RV S’: β -145.1 (-260.7 to -29.6), p=0.015]. Conclusions: In this cohort of children with sepsis, echocardiographic indices as measured by our point of care, novel algorithms can be utilized to risk stratify these high-risk patients. Right sided chambers appear to be more predictive of both mortality and morbidity.

Operational and Analytical Modal Analysis of a Bridge Using Low-Cost Wireless Arduino-Based Accelerometers.

Arduino-based accelerometers are receiving wide attention from researchers to make long-term Structural Health Monitoring (SHM) feasible for structures with a low SHM budget. The current low-cost solutions found in the literature share some of the following drawbacks: (1) high noise density, (2) lack of wireless synchronization, (3) lack of automatic data acquisition and data management, and (4) lack of dedicated field tests aiming to compare mode shapes from Operational Modal Analysis (OMA) with those of a digital model. To solve these problems, a recently built short-span footbridge in Barcelona is instrumented using four Low-cost Adaptable Reliable Accelerometers (LARA). In this study, the automatization of the data acquisition and management of these low-cost solutions is studied for the first time in the literature. In addition, a digital model of the bridge under study is generated in SAP2000 using the available drawings and reported characteristics of its materials. The OMA of the bridge is calculated using Frequency Domain Decomposition (FDD) and Covariance Stochastic Subspace Identification (SSI-cov) methods. Using the Modal Assurance Criterion (MAC), the mode shapes of OMA are compared with those of the digital model. Finally, the acquired eigenfrequencies of the bridge obtained with a high-precision commercial sensor (HI-INC) showed a good agreement with those obtained with LARA.

XPlaM: A toolkit for automating the acquisition of BDI agent-based Digital Twins of organizations

A Digital Twin ideally manifests the same behaviour (in silico) of its physical counterpart. While considerable attention has been paid to the development of Digital Twins for physical devices/systems, the question of developing Twins for organizations has received relatively little attention. The setting in which we address this problem is very general and, consequently, very challenging. We look at the automatic acquisition of Digital Twins of organizations. To that end, this paper builds on the following two premises: (1) that Digital Twins of organizations can provide value, and (2) that Belief-Desire-Intention (BDI) agents are a particularly effective means for representing Digital Twins. The overall approach is to leverage the externally observable behaviour of the target system and then generate candidate BDI agent programs that best explain (in the sense of formal abduction) the observed behaviour. The candidate agent programs are generated by searching through potentially large hypotheses spaces for possible plans, selection functions and beliefs. The resulting approach suggests that using abduction to generate Digital Twins of organizations in the form of BDI agents can be effective.

Apple detection and instance segmentation in natural environments using an improved Mask Scoring R-CNN Model.

The accurate detection and segmentation of apples during growth stage is essential for yield estimation, timely harvesting, and retrieving growth information. However, factors such as the uncertain illumination, overlaps and occlusions of apples, homochromatic background and the gradual change in the ground color of apples from green to red, bring great challenges to the detection and segmentation of apples. To solve these problems, this study proposed an improved Mask Scoring region-based convolutional neural network (Mask Scoring R-CNN), known as MS-ADS, for accurate apple detection and instance segmentation in a natural environment. First, the ResNeSt, a variant of ResNet, combined with a feature pyramid network was used as backbone network to improve the feature extraction ability. Second, high-level architectures including R-CNN head and mask head were modified to improve the utilization of high-level features. Convolutional layers were added to the original R-CNN head to improve the accuracy of bounding box detection (bbox_mAP), and the Dual Attention Network was added to the original mask head to improve the accuracy of instance segmentation (mask_mAP). The experimental results showed that the proposed MS-ADS model effectively detected and segmented apples under various conditions, such as apples occluded by branches, leaves and other apples, apples with different ground colors and shadows, and apples divided into parts by branches and petioles. The recall, precision, false detection rate, and F1 score were 97.4%, 96.5%, 3.5%, and 96.9%, respectively. A bbox_mAP and mask_mAP of 0.932 and 0.920, respectively, were achieved on the test set, and the average run-time was 0.27 s per image. The experimental results indicated that the MS-ADS method detected and segmented apples in the orchard robustly and accurately with real-time performance. This study lays a foundation for follow-up work, such as yield estimation, harvesting, and automatic and long-term acquisition of apple growth information.

Automatic acquisition method for evaluating the on-orbit modulation transfer function based on lunar targets.

In view of the existing method having a great subjectivity for the lunar edge selection, we propose an automatic knife-edge detection method based on the Hough transform to evaluate the on-orbit modulation transfer function (MTF) of the satellite remote sensor. This novel method avoids the dependence of the on-orbit MTF measurement on the edge selection location, overcoming the limitation of the traditional method needing to identify the lunar shape and fix its edge area. On basis of two different edge detection algorithms (Sobel operator and Prewitt operator), the binary edge images of the moon are acquired, thus obtaining a series of edges satisfying the determination requirement by the Hough transform, and the MTFs corresponding to each knife-edge are calculated to obtain the optimal MTF. The automatic knife-edge detection method greatly improves the accuracy of the lunar edge selection, and the MTF obtained by the novel method is obviously better than that of the traditional method. In order to verify the effectiveness of the novel method, the long time series of the on-orbit MTFs for the FY-2G and FY-2E satellite measurements are given, indicating that the FY-2E observation has higher stability and better performance compared with that of the FY-2G satellite. This study has an important practical significance for evaluating the on-orbit stability of the satellite its optical imaging quality.

Development and application of the multifunctional transmission roll-sliding friction tester

This paper aims to elaborate on the development and the main features of the multifunctional transmission roll-sliding friction tester. The tester adopts the independent drive for controlled relative movement, and the contacting specimen that sustains the radial loads can be replaced, so it can complete the rolling friction, sliding friction, and roll-sliding friction test. The integrated transmission performance test machine not only has solved the inconsistency between the friction and transmission experimental boundary conditions but also can precede the friction drive research. Piecewise calibration data and interpolating correction algorithm are used to avoid the affecting generated by the spindle friction to the specimen. The monitoring system software is developed by the LabVIEW platform, and the parameters are controlled automatically by the computer. Thus it can simulate actual working conditions very well and achieve automatic real-time data acquisition, processing and showing the results, etc.

Lightning Whistler Wave Speech Recognition Based on Grey Wolf Optimization Algorithm

The recognition algorithm of the lightning whistler wave, based on intelligent speech, is the key technology to break the bottleneck of massive data and study the temporal and spatial variation rules of the lightning whistler wave. However, its recognition effect depends on the hyperparameters determined by manual experiments repeatedly, which takes a great deal of time and cannot guarantee the best recognition effect of the model. Therefore, we proposed the lightning whistler wave recognition algorithm based on grey wolf optimization (GWO). In this paper, the GWO algorithm is used to automatically find the best value of hyperparameters of Long Short-Term Memory (LSTM) in their limited searching space. Here we consider the number of hidden units (hu) and learning rate (lr) as the hyperparameters to be optimized, and the spatial coordinate (hu, lr) as the grey wolf position. By the end of the GWO process, we obtain the position of the wolf king α with the optimal hu and lr searched by the GWO algorithm. Then we use the optimal hu and lr to configure LSTM and perform supervised learning on the train set to obtain the final lightning whistler wave speech recognition model. Through experimental verification, the recognition model based on the GWO not only overcomes the uncertainty of the traditional model relying on manual finetuning of parameters and realizes the mechanism of automatic search and acquisition of hyperparameters, but also its recognition effect improves by about 2% in accuracy, F1score, and other metrics compared with the model trained by manually setting hyperparameters.

Research on Autopilot Trajectory Planning Based on Deep Learning

Vehicle trajectory is usually used in driving behavior modeling and traffic safety. In order to obtain vehicle tracks in traffic video, an automatic acquisition method based on depth learning is proposed. In this paper, the YOLOV5 detector is used to detect vehicles in the region of interest in continuous frames, and the DEEPSORT tracker is used to track robust and fast vehicles. This tracker can effectively reduce the jump of target ID, improve tracking stability, obtain the real-world coordinates of vehicles through perspective transformation, and use a locally weighted regression algorithm to smooth the track. The experimental results show that the average multi-target tracking accuracy of the method in the expressway monitoring data set is 96.1%, and the average accuracy of trajectory acquisition is more than 93%. The smoothing algorithm can effectively remove the trajectory noise.

Detecting strawberry diseases and pest infections in the very early stage with an ensemble deep-learning model.

Detecting early signs of plant diseases and pests is important to preclude their progress and minimize the damages caused by them. Many methods are developed to catch signs of diseases and pests from plant images with deep learning techniques, however, detecting early signs is still challenging because of the lack of datasets to train subtle changes in plants. To solve these challenges, we built an automatic data acquisition system for the accumulation of a large dataset of plant images and trained an ensemble model to detect targeted plant diseases and pests. After obtaining 13,393 plant image data, our ensemble model shows a decent detection performance with an average of AUPRC 0.81. Also, this data acquisition and the detection process can be applied to other plant anomalies with the collection of additional data.

Automatic Spoken Language Acquisition Based on Observation and Dialogue

Human babies are born without knowledge of any specific language. They acquire language directly from observation and dialogue without being limited by the availability of labeled data. We propose spoken language acquisition agents that simulate the process. Such an ability requires multiple types of learning, including 1) word discovery, 2) symbol grounding, 3) message generation, and 4) pronunciation generation. Several studies have targeted one or combined learning types to elucidate human intelligence and aimed to equip spoken dialogue systems with human-like flexible language learning ability. However, their language ability was partially lacking some of the components. Our agents are the first to integrate them all. Our key concept is to design an architecture to integrate unsupervised, self-supervised, and reinforcement learning to utilize clues naturally existing in raw sensory signals and drive the learning based on the agent’s intrinsic motivation. Experimental results show agents successfully acquire spoken language from scratch by interacting with an environment to act by speaking. Our proposed focusing mechanism significantly improves learning efficiency. We also demonstrate that our agents can learn neural vocoder and the concept of logical negation as a part of language acquisition.

End-to-End Plankton Database Collection System

Monitoring living plankton status has greatly developed owing to machine vision technology. One of these developments is the automatic video acquisition system, which can capture high-resolution information and record considerable details of a scene. However, existing optical sensors cannot usually obtain a whole image, in which all details of plankton are fully clear, due to the influence of many factors, such as depth-of-field limitation of lenses, movement of plankton, and large difference in plankton scale. The captured video also needs to be filtered to eliminate the redundancy before using for the sparsity of the sample. This procedure is time-consuming and costly. Therefore, in this article, we develop an end-to-end plankton database collection system that can directly generate complete, clear plankton images from video. First, the regions of interest of the plankton are extracted. Then, the same plankton which appears in successive frames is identified to reconstruct the clearest morphological and structure features through their detailed information. Experimental results indicate that the proposed system can effectively compress the original data. The proposed fusion method also outperforms the state-of-the-art methods, especially for images with anisotropic blur. Furthermore, this system can monitor the abundance and distribution of marine plankton with the help of an embedded computing platform.

A Self-Supervised Learning-Based 6-DOF Grasp Planning Method for Manipulator

To realize a robust robotic grasping system for unknown objects in an unstructured environment, large amounts of grasp data and 3D model data for the object are required; the sizes of these data directly affect the rate of successful grasps. To reduce the time cost of data acquisition and labeling and increase the rate of successful grasps, we developed a self-supervised learning mechanism to control grasp tasks performed by manipulators. First, a manipulator automatically collects the point cloud for the objects from multiple perspectives to increase the efficiency of data acquisition. The complete point cloud for the objects is obtained using the hand-eye vision of the manipulator and the truncated signed distance function algorithm. Then, the point cloud data for the objects are used to generate a series of six-degrees-of-freedom grasp poses, and the force-closure decision algorithm is used to add the grasp quality label to each grasp pose to realize the automatic labeling of grasp data. Finally, the point cloud in the gripper closing area corresponding to each grasp pose is obtained and used to train the grasp-quality classification model for the manipulator. The results of performing actual grasping experiments demonstrate that the proposed self-supervised learning method can increase the rate of successful grasps for the manipulator. Note to Practitioners—Most of the existing grasp planning methods of the manipulator are based on public datasets or simulation data to train model algorithms. Owing to the limited types of objects, the limited amount of data in the public datasets, and the lack of real sensor noise in the simulation data, the robustness of the trained algorithm model is insufficient, and it is difficult to apply to unstructured production environments. To solve the above problems, we propose a 6-DOF capture planning method based on self-supervised learning and introduce a self-supervised learning mechanism to solve the problem of grasp data acquisition in real scenes. The manipulator automatically collects object data from multiple perspectives, performs desktop-level 3D reconstruction, and finally uses the force-closure decision algorithm to automatically label the data in order to realize automatic acquisition and labeling of the grasp data in a real scenario. Preliminary experiments show that this method can obtain high-quality grasp data and can be applied to grasp operations in real multi-target and cluttered environments. However, it has not been tested in actual production environments. This paper focuses on the data acquisition module in the 6-DOF grasp planning framework. In future research, we will design a more efficient grasp planning module to improve the grasp efficiency of the manipulator.

Superpixel segmentations for thin sections: Evaluation of methods to enable the generation of machine learning training data sets

Training data is the backbone of developing either Machine Learning (ML) models or specific Deep Learning (DL) algorithms. The lack of well-labeled training image data has significantly impeded the development of novel DL methods like Convolutional Neural Networks (CNNs) in mineral thin section images identification. However, image annotation, especially pixel-wise annotation is always a costly process. Manually creating dense semantic labels for rock thin section images has been long considered as an unprecedented challenge in view of the ubiquitous variety and complexity of minerals in thin sections. To speed up the annotation, we propose a human–computer collaborative pipeline in which superpixel segmentation is used as a boundary extractor to avoid hand delineation of instances boundaries. The pipeline consists of two steps: superpixel segmentation using MultiSLIC, and superpixel labeling through a specific-designed tool. We use a cutting-edge methodology Virtual Petroscopy (ViP) for automatic image acquisition. Bentheimer sandstone sample is used to conduct performance testing of the pipeline. Three standard error metrics are used to evaluate the performance of MultiSLIC. The result indicates that MultiSLIC is able to extract compact superpixels with satisfying boundary adherence given multiple input images. According to our test results, large and complex thin section images with pixel-wise accurate labels can be annotated with the labeling tool more efficiently than in a conventional, purely manual work, and generate data of high quality.