Set Of Images Research Articles

A Proof-of-Concept Computer Vision Approach for Measurement of Tympanic Membrane Perforations.

Size, an important characteristic of a tympanic membrane perforation (TMP), is commonly assessed with gross estimation via visual inspection, a practice which is prone to inaccuracy. Herein, we demonstrate feasibility of a proof-of-concept computer vision model for estimating TMP size in a small set of perforations. An open-source deep learning architecture was used to train a model to segment and calculate the area of a perforation and the visualized tympanic membrane (TM) in a set of endoscopic images of mostly anterior and relatively small TMPs. The model then computed relative TMP size by calculating the ratio of perforation area to TM area. Model performance on the test dataset was compared to ground-truth manual annotations. In a validation survey, otolaryngologists were tasked with estimating the size of TMPs from the test dataset. The primary outcome was the average absolute error of model size predictions and clinician estimates compared to sizes determined by ground-truth manual annotations. The model's average absolute error for size predictions was a 0.8% overestimation for all test perforations. Conversely, among the 38 survey respondents, the average clinician error was a 11.0% overestimation (95% CI, 5.2-16.7%, p = 0.003). In a small sample of TMPs, we demonstrated a computer vision approach for estimating TMP size is feasible. Further validation studies must be done with significantly larger and more heterogenous datasets. N/A Laryngoscope, 134:2906-2911, 2024.

The Impact of Closed-Loop Imaging on Actionable CT-Detected Breast Findings

PurposeClosed-loop imaging programs (CLIPs) are designed to ensure that patients receive appropriate follow-up, but a review of incidental CT-detected breast findings in the setting of CLIPs has not been performed. MethodsA retrospective review was conducted of CT reports at a single academic institution from July 1, 2020, to January 31, 2022, to identify reports with recommendations for breast imaging follow-up. Medical records were reviewed to evaluate patient adherence to follow-up, CLIP intervention, subsequent BI-RADS assessment, and diagnosis. Adherence was defined as diagnostic breast imaging performed within 6 months of the CT recommendation. ResultsFollow-up recommendations for breast imaging were included in CT report impressions for 311 patients. Almost half of patients (47.3% [147 of 311]) underwent follow-up breast imaging within 6 months, yielding breast cancer diagnoses in 12.9% (19 of 147) and a biopsy-proven positive predictive value of 65.5% (19 of 29). Most patients who returned for follow-up within 6 months did so without CLIP intervention. The majority of CT report impressions in the follow-up group (85.0% [125 of 147]) contained specific recommendations for “diagnostic breast imaging.” For patients who did not receive follow-up, the CLIP team tracked all cases and intervened in 19.1% (28 of 147). The most common intervention was a phone call and/or fax to the primary care provider. Outpatient CT examination setting and specific recommendation for diagnostic breast imaging were significantly associated with higher follow-up adherence (P < .0001). ConclusionsActionable CT-detected breast findings require follow-up diagnostic breast imaging because of a relevant cancer detection rate of 12.9%. Although many patients return for breast imaging without intervention, almost half of patients did not receive follow-up and may account for a significant number of missed cancer diagnoses. Specific CT recommendation verbiage is associated with higher follow-up adherence, which can be addressed across settings even without CLIPs.

PS-NET: an end-to-end phase space depth estimation approach for computer-generated holograms.

In the present work, an end-to-end approach is proposed for recovering an RGB-D scene representation directly from a hologram using its phase space representation. The proposed method involves four steps. First, a set of silhouette images is extracted from the hologram phase space representation. Second, a minimal 3D volume that describes these silhouettes is extracted. Third, the extracted 3D volume is decomposed into horizontal slices, and each slice is processed using a neural network to generate a coarse estimation of the scene geometry. Finally, a third neural network is employed to refine the estimation for higher precision applications. Experimental results demonstrate that the proposed approach yields faster and more accurate results compared to numerical reconstruction-based methods. Moreover, the obtained RGB-D representation can be directly utilized for alternative applications such as motion estimation.

Principal Component Approximation Network for Image Compression

In this work, we propose a novel principal component approximation network (PCANet) for image compression. The proposed network is based on the assumption that a set of images can be decomposed into several shared feature matrices, and an image can be reconstructed by the weighted sum of these matrices. The proposed PCANet is specifically devised to learn and approximate these feature matrices and weight vectors, which are used to encode images for compression. Unlike previous deep learning-based methods, a distinctive aspect of our approach is its consideration of network size in the bit-rate computation. Despite this inclusion, our proposed method yields promising results. Through extensive experiments conducted on standard datasets, we demonstrate the effectiveness of our approach in comparison to state-of-the-art techniques. To the best of our knowledge, this is the first machine learning approach that includes the size of networks during bitrate computation in image compression.

Comparative Assessment of Neural Radiance Fields and Photogrammetry in Digital Heritage: Impact of Varying Image Conditions on 3D Reconstruction

This paper conducts a comparative evaluation between Neural Radiance Fields (NeRF) and photogrammetry for 3D reconstruction in the cultural heritage domain. Focusing on three case studies, of which the Terpsichore statue serves as a pilot case, the research assesses the quality, consistency, and efficiency of both methods. The results indicate that, under conditions of reduced input data or lower resolution, NeRF outperforms photogrammetry in preserving completeness and material description for the same set of input images (with known camera poses). The study recommends NeRF for scenarios requiring extensive area mapping with limited images, particularly in emergency situations. Despite NeRF’s developmental stage compared to photogrammetry, the findings demonstrate higher potential for describing material characteristics and rendering homogeneous textures with enhanced visual fidelity and accuracy; however, NeRF seems more prone to noise effects. The paper advocates for the future integration of NeRF with photogrammetry to address respective limitations, offering more comprehensive representation for cultural heritage preservation tasks. Future developments include extending applications to planar surfaces and exploring NeRF in virtual and augmented reality, as well as studying NeRF evolution in line with emerging trends in semantic segmentation and in-the-wild scene reconstruction.

SatellStitch: Satellite Imagery-Assisted UAV Image Seamless Stitching for Emergency Response without GCP and GNSS

Rapidly stitching unmanned aerial vehicle (UAV) imagery to produce high-resolution fast-stitch maps is key to UAV emergency mapping. However, common problems such as gaps and ghosting in image stitching remain challenging and directly affect the visual interpretation value of the imagery product. Inspired by the data characteristics of high-precision satellite images with rich access and geographic coordinates, a seamless stitching method is proposed for emergency response without the support of ground control points (CGPs) and global navigation satellite systems (GNSS). This method aims to eliminate stitching traces and solve the problem of stitching error accumulation. Firstly, satellite images are introduced to support image alignment and geographic coordinate acquisition simultaneously using matching relationships. Then a dynamic contour point set is constructed to locate the stitching region and adaptively extract the fused region of interest (FROI). Finally, the gradient weight cost map of the FROI image is computed and the Laplacian pyramid fusion rule is improved to achieve seamless production of the fast-stitch image map with geolocation information. Experimental results indicate that the method is well adapted to two representative sets of UAV images. Compared with the Laplacian pyramid fusion algorithm, the peak signal-to-noise ratio (PSNR) of the image stitching results can be improved by 31.73% on average, and the mutual information (MI) can be improved by 19.98% on average. With no reliance on CGPs or GNSS support, fast-stitch image maps are more robust in harsh environments, making them ideal for emergency mapping and security applications.

Siamese model for collateral score prediction from computed tomography angiography images in acute ischemic stroke.

Imaging biomarkers, such as the collateral score as determined from Computed Tomography Angiography (CTA) images, play a role in treatment decision making for acute stroke patients. In this manuscript, we present an end-to-end learning approach for automatic determination of a collateral score from a CTA image. Our aim was to investigate whether such end-to-end learning approaches can be used for this classification task, and whether the resulting classification can be used in existing outcome prediction models. The method consists of a preprocessing step, where the CTA image is aligned to an atlas and divided in the two hemispheres: the affected side and the healthy side. Subsequently, a VoxResNet based convolutional neural network is used to extract features at various resolutions from the input images. This is done by using a Siamese model, such that the classification is driven by the comparison between the affected and healthy using a unique set of features for both hemispheres. After masking the resulting features for both sides with the vascular region and global average pooling (per hemisphere) and concatenation of the resulting features, a fully connected layer is used to determine the categorized collateral score. Several experiments have been performed to optimize the model hyperparameters and training procedure, and to validate the final model performance. The hyperparameter optimization and subsequent model training was done using CTA images from the MR CLEAN Registry, a Dutch multi-center multi-vendor registry of acute stroke patients that underwent endovascular treatment. A separate set of images, from the MR CLEAN Trial, served as an external validation set, where collateral scoring was assessed and compared with both human observers and a recent more traditional model. In addition, the automated collateral scores have been used in an existing functional outcome prediction model that uses both imaging and non-imaging clinical parameters. The results show that end-to-end learning of collateral scoring in CTA images is feasible, and does perform similar to more traditional methods, and the performance also is within the inter-observer variation. Furthermore, the results demonstrate that the end-to-end classification results also can be used in an existing functional outcome prediction model.

Brain activation during scene encoding fMRI in the Alzheimer’s disease continuum: Association with amyloid and tau burden in PE

Background and Hypothesis:This project assessed brain activation during a scene encoding task in 4 groups: older adults who were cognitively normal (CN), subjective cognitive decline (SCD), mild cognitive impairment (MCI), and dementia due to Alzheimer’s disease (AD). Associations between scene encoding related brain activation and tau, amyloid, and other biomarkers were analyzed. Our hypothesis was that higher levels of cerebral tau and amyloid would be associated with reduced scene encoding activation. In addition, we hypothesized that scene encoding activation would be significantly different between cognitively normal and cognitively impaired groups.&#x0D; Methods:234 individuals from the Indiana Memory and Aging Study (79 CN, 67 SCD, 70 MCI, and 18 AD) completed structural and functional MRI, clinical/cognitive assessment and biomarkers; 155 underwent amyloid ([18F]florbetapir/[18F]florbetaben) PET, while 111 also underwent [18F]flortaucipir PET. For the fMRI scene encoding task, participants were asked to view and remember a set of images. A one-way ANOVA test was used to analyze scene encoding related activation differences among the 4 groups. Regression was used to identify associations between scene encoding activation and tau and amyloid deposition.&#x0D; Results:Significant differences in activation were observed between the MCI and CN groups, including less activation in widespread regions during the task and reduced deactivation in the default mode network (DMN) in MCI participants relative to CN. Significant associations between higher amyloid and tau deposition and altered scene encoding activation were also observed.&#x0D; Conclusion and Potential Impact:Cognitive decline is associated with activation changes during scene encoding, as well as reduced deactivation in the DMN, especially in the posterior cingulate region. Higher cerebral amyloid and tau deposition predicted decreased scene encoding related activation. These findings are consistent with models linking cognitive status, functional brain activation during episodic encoding, and pathophysiological processes in the AD continuum.

Holistic Processing of Recognition Memory for Face Identity: Evidence from a Face Morphing Paradigm

AbstractHolistic face processing refers to a tendency to see face parts as a whole. This study examined whether and how holistic processing affects recognition memory for face identity in a learned/novel morphing continuum. In the study phase of each experiment reported here, adult participants were asked to memorize (i.e., learn) a set of unfamiliar face images by viewing them once or twice. In the test phase, the participants were presented with face images morphed from learned and novel ones and were asked to report whether they viewed the face in the study phase. In Experiment 1, we demonstrated that our morphing paradigm can capture memory selectivity changes (i.e., recognition boundary and sensitivity) associated with learned face identity. In Experiment 2, we demonstrated that whole‐face presentations improved sensitivity to learned/novel face identity, but did not alter the recognition boundary between learned and novel face identity. These results support the view that holistic processing contributes to accurate face identification in recognition memory, even under minimal learning situations.

Deep learning application for abdominal organs segmentation on 0.35 T MR-Linac images.

Linear accelerator (linac) incorporating a magnetic resonance (MR) imaging device providing enhanced soft tissue contrast is particularly suited for abdominal radiation therapy. In particular, accurate segmentation for abdominal tumors and organs at risk (OARs) required for the treatment planning is becoming possible. Currently, this segmentation is performed manually by radiation oncologists. This process is very time consuming and subject to inter and intra operator variabilities. In this work, deep learning based automatic segmentation solutions were investigated for abdominal OARs on 0.35 T MR-images. One hundred and twenty one sets of abdominal MR images and their corresponding ground truth segmentations were collected and used for this work. The OARs of interest included the liver, the kidneys, the spinal cord, the stomach and the duodenum. Several UNet based models have been trained in 2D (the Classical UNet, the ResAttention UNet, the EfficientNet UNet, and the nnUNet). The best model was then trained with a 3D strategy in order to investigate possible improvements. Geometrical metrics such as Dice Similarity Coefficient (DSC), Intersection over Union (IoU), Hausdorff Distance (HD) and analysis of the calculated volumes (thanks to Bland-Altman plot) were performed to evaluate the results. The nnUNet trained in 3D mode achieved the best performance, with DSC scores for the liver, the kidneys, the spinal cord, the stomach, and the duodenum of 0.96 ± 0.01, 0.91 ± 0.02, 0.91 ± 0.01, 0.83 ± 0.10, and 0.69 ± 0.15, respectively. The matching IoU scores were 0.92 ± 0.01, 0.84 ± 0.04, 0.84 ± 0.02, 0.54 ± 0.16 and 0.72 ± 0.13. The corresponding HD scores were 13.0 ± 6.0mm, 16.0 ± 6.6mm, 3.3 ± 0.7mm, 35.0 ± 33.0mm, and 42.0 ± 24.0mm. The analysis of the calculated volumes followed the same behavior. Although the segmentation results for the duodenum were not optimal, these findings imply a potential clinical application of the 3D nnUNet model for the segmentation of abdominal OARs for images from 0.35 T MR-Linac.

Radiation source personalization for nanoparticle-enhanced radiotherapy using dynamic contrast-enhanced MRI in the treatment planning process

Nanoparticle-enhanced radiotherapy offers the potential to selectively increase the radiation dose imparted to the tumor while at the same time sparing the healthy structures around it. Among the recommendations of an interdisciplinary group for the clinical translation of this treatment modality, is the developing of methods to quantify the effects that the nanoparticle concentration has on the radiation dosimetry and incorporate these effects into the treatment planning process. In this work, using Monte Carlo simulations and dynamic contrast-enhanced MRI images, treatment plans for nanoparticle-enhanced radiotherapy are calculated in order to evaluate the effects that realistic distributions of the nanoparticles have on the resultant plans and to devise treatment strategies to account for these effects, including the selection of the proper x-ray source configuration in terms of energy and collimation. Dynamic contrast-enhanced MRI studies were obtained for two treatment sites, namely brain and breast. A model to convert the MRI signal to contrast agent concentration was applied to each set of images. Two treatment modalities, 3D conformal radiotherapy and Stereotactic Body Radiation Therapy, were evaluated at three different x-ray spectra, namely 6 MV from a linear accelerator, 110 kVp and 220 kVp from a tungsten target. For the breast patient, as the nanoparticle distribution varies markedly with time, the treatment plans were obtained at two different times after administration. It was determined that maximum doses to the healthy structures around the tumor are mostly determined by the minimum nanoparticle concentration in the tumor. The presence of highly hypoxic or necrotic tissue, which fails to accumulate the nanoparticles, or leakage of the contrast agent into the surrounding healthy tissue, make irradiation with conventional conformal radiotherapy unfeasible for kilovoltage beam energies, as the uniform beam apertures lack the ability to compensate for the non-uniform distribution of the nanoparticles. Therefore, proper quantification of the nanoparticle distribution not only in the target volume but also in the surrounding tissues and structures is crucial for the proper planning of nanoparticle-enhanced radiotherapy and a treatment delivery with a high-degree of freedom, such as small-field stereotactic body radiotherapy, should be the method of choice for this treatment modality.

Deep learning segmentation of endothelial cell images using an active learning paradigm with guided label corrections.

To create Guided Correction Software for informed manual editing of automatically generated corneal endothelial cell (EC) segmentations and apply it to an active learning paradigm to analyze a diverse set of post-keratoplasty EC images. An original U-Net model trained on 130 manually labeled post-Descemet stripping automated endothelial keratoplasty (EK) images was applied to 841 post-Descemet membrane EK images generating "uncorrected" cell border segmentations. Segmentations were then manually edited using the Guided Correction Software to create corrected labels. This dataset was split into 741 training and 100 testing EC images. U-Net and DeepLabV3+ were trained on the EC images and the corresponding uncorrected and corrected labels. Model performance was evaluated in a cell-by-cell analysis. Evaluation metrics included the number of over-segmentations, under-segmentations, correctly identified new cells, and endothelial cell density (ECD). Utilizing corrected segmentations for training U-Net and DeepLabV3+ improved their performance. The average number of over- and under-segmentations per image was reduced from 23 to 11 with the corrected training set. Predicted ECD values generated by networks trained on the corrected labels were not significantly different than the ground truth counterparts (, paired -test). These models also correctly segmented a larger percentage of newly identified cells. The proposed Guided Correction Software and semi-automated approach reduced the time to accurately segment EC images from 15 to 30 to 5min, an decrease compared to manual editing. Guided Correction Software can efficiently label new training data for improved deep learning performance and generalization between EC datasets.

Enhanced video clustering using multiple riemannian manifold-valued descriptors and audio-visual information

Videos inherently blend multiple modalities in real-world scenarios, primarily visual and auditory cues. When synergized, these cues foster enhanced data representations. Standard clustering techniques, primarily designed for managing vectorial data in Euclidean spaces, struggle to handle multidimensional data with nonlinear manifold structures, such as video or image sets. While recent subspace clustering methods using Riemannian manifold representation tackle this issue, they often sideline auditory information, overlooking the potential harmony between visual and auditory modalities. This paper presents an innovative approach that crafts multiple Riemannian manifold-valued descriptors to bridge this gap, encapsulating multimodal video information in a unified structure. We architect a single-modality Riemannian subspace clustering for individual modal data and extend it to a multi-modality framework, leveraging the interplay of audio-visual data. Detailed optimization and convergence analysis are also provided. The proposed approach significantly outperforms the existing state-of-the-art methods, improving accuracy by 4%, 1%, and 2% on UCF-101, UCF-sport, and AVE datasets, respectively.

A multiscale feature fusion enhanced CNN with the multiscale channel attention mechanism for efficient landslide detection (MS2LandsNet) using medium-resolution remote sensing data

ABSTRACT Deep learning (DL) models have been widely used for remote sensing-based landslide mapping due to their impressive capabilities for automatic information extraction. However, the large volumes of parameters and calculations have compromised the efficiency of DL models in extracting landslides from a large set of RS images. Lightweight convolutional neural networks (CNNs) exhibit promising feature representation abilities with fewer parameters. This study aims to introduce a new lightweight CNN called MS2LandsNet, designed to detect landslides with both high efficiency and accuracy. The MS2LandsNet consists of three down-sampling stages embedded with multi-scale feature fusion (MFF), aiming to decrease parameters while aggregating contextual features. Additionally, we incorporate multi-scale channel attention (MSCA) into MFF to improve performance. According to experimental results on three landslip datasets, MS2LandsNet obtains the highest F1 score of 85.90% and the highest IoU of 75.28%. Notably, MS2LandsNet accomplishes the resuts with the fewest parameters and the fastest inference speed, outperforming seven classical semantic segmentation models and three lightweight CNNs. The proposed lightweight model holds potential for application on a cloud computing platform for larger-scale landslide mapping tasks in future work.

Neural network approach for shape-based euhedral pyrite identification in X-ray CT data with adversarial unsupervised domain adaptation

We explore an attenuation and shape-based identification of euhedral pyrites in high-resolution X-ray Computed Tomography (XCT) data using deep neural networks. To deal with the scarcity of annotated data we generate a complementary training set of synthetic images. To investigate and address the domain gap between the synthetic and XCT data, several deep learning models, with and without domain adaption, are trained and compared. We find that a model trained on a small set of human annotations, while displaying over-fitting, can rival the human annotators. The unsupervised domain adaptation approaches are successful in bridging the domain gap, which significantly improves their performance. A domain-adapted model, trained on a dataset that fuses synthetic and real data, is the overall best-performing model. This highlights the possibility of using synthetic datasets for the application of deep learning in mineralogy.

New insights into multi-focus image fusion: A fusion method based on multi-dictionary linear sparse representation and region fusion model

This paper proposes a method based on multi-dictionary linear sparse representation and region fusion model for achieving high-quality multi-focus image fusion (MFIF). First, a multi-dictionary linear sparse representation model (MDLSR) is proposed for focus detection of source images, in which the focus information of the source image is characterized using linear coefficients between the dictionary and the input image, while the dictionary is a set of images including the source images. MDLSR exploits the correlation of feature information among local neighborhoods of the source image, and this feature significantly improves the accuracy of focus detection. Next, we propose an edge-weighting closed-form matting (EWCM) to define the source image focus boundary accurately. EWCM calculates the weights at each position in a content-adaptive manner, suppressing the effects of vignetting artifacts near the edges to preserve more edge details. Then, to avoid the focused/defocused boundary (FDB) region fusion problem caused by the defocus spread effect (DSE), we propose a region fusion model (RFM) for the MFIF, which uses correction and compensation to mitigate this problem. Finally, we also adopt an iterative update approach to improve the quality of the fused images. These designs achieve a win-win situation regarding fusion performance and realization ability. The experimental results on four typical datasets show that our method outperforms 37 state-of-the-art methods in terms of subjective visual effect, objective metric analysis and generalization ability. Our code is available at https://github.com/JWatawang/MDLSR_RFM.

Levator Scapulae Stiffness Measurement Reliability in Individuals with and without Chronic Neck Pain by Experienced and Novel Examiners.

The levator scapulae muscle is a key structure in the etiopathology of neck and shoulder musculoskeletal pain. Although previous studies used shear-wave elastography (SWE) for characterizing this muscle elasticity, limited evidence assessed the inter-examiner reliability of this procedure. This study aimed to analyze the inter-examiner reliability for calculating Young's modulus and shear wave speed in a cohort of participants with and without chronic neck pain. A diagnostic accuracy study was conducted, acquiring a set of SWE images at the C5 level in participants with and without neck pain (n = 34 and 33, respectively) by two examiners (one experienced and one novel). After blinding the participants' identity, examiner involved, and side, the stiffness indicators were calculated by an independent rater in a randomized order. Intra-class correlation coefficients (ICC), standard error of measurement, minimal detectable changes, and coefficient of variation were calculated. Both cohorts had comparable sociodemographic characteristics (p > 0.05). No significant levator scapulae elasticity differences were found between genders, sides, or cohorts (all, p > 0.05). Inter-examiner reliability for calculating Young's modulus and shear wave speed was moderate-to-good for assessing asymptomatic individuals (ICC = 0.714 and 0.779, respectively), while poor-to-moderate in patients with neck pain (ICC = 0.461 and 0.546, respectively). The results obtained in this study support the use of this procedure for assessing asymptomatic individuals. However, reliability estimates were unacceptable to support its use for assessing elasticity in patients with chronic neck pain. Future studies might consider that the shear wave speed is more sensitive to detect real changes in comparison with Young's modulus.

Comprehensive advancement in endoscopy: optical design, algorithm enhancement, and clinical validation for merged WLI and CBI imaging.

As endoscopic imaging technology advances, there is a growing clinical demand for enhanced imaging capabilities. Although conventional white light imaging (WLI) endoscopy offers realistic images, it often cannot reveal detailed characteristics of the mucosa. On the other hand, optical staining endoscopy, such as Compound Band Imaging (CBI), can discern subtle structures, serving to some extent as an optical biopsy. However, its image brightness is low, and the colors can be abrupt. These two techniques, commonly used in clinical settings, have complementary advantages. Nonetheless, they require different lighting conditions, which makes it challenging to combine their imaging strengths on living tissues. In this study, we introduce a novel endoscopic imaging technique that effectively combines the advantages of both WLI and CBI. Doctors don't need to manually switch between these two observation modes, as they can obtain the image information of both modes in one image. We calibrated an appropriate proportion for simultaneous illumination with the light required for WLI and CBI. We designed a new illumination spectrum tailored for gastrointestinal examination, achieving their fusion at the optical level. Using a new algorithm that focuses on enhancing specific hemoglobin tissue features, we restored narrow-band image characteristics lost due to the introduction of white light. Our hardware and software innovations not only boost the illumination brightness of the endoscope but also ensure the narrow-band feature details of the image. To evaluate the reliability and safety of the new endoscopic system, we conducted a series of tests in line with relevant international standards and validated the design parameters. For clinical trials, we collected a total of 256 sets of images, each set comprising images of the same lesion location captured using WLI, CBI, and our proposed method. We recruited four experienced clinicians to conduct subjective evaluations of the collected images. The results affirmed the significant advantages of our method. We believe that the novel endoscopic system we introduced has vast potential for clinical application in the future.

Preparing multi-layered visualisations of Old Babylonian cuneiform tablets for a machine learning OCR training model towards automated sign recognition

Abstract In the framework of the CUNE-IIIF-ORM project the aim is to train an Artificial Intelligence Optical Character Recognition (AI-OCR) model that can automatically locate and identify cuneiform signs on photorealistic representations of Old Babylonian texts (c. 2000–1600 B.C.E.). In order to train the model, c. 200 documentary clay tablets have been selected. They are manually annotated by specialist cuneiformists on a set of 12 still raster images generated from interactive Multi-Light Reflectance images. This image set includes visualisations with varying light angles and simplifications based on the dept information on the impressed signs in the surface. In the Cuneur Cuneiform Annotator, a Gitlab-based web application, the identified cuneiform signs are annotated with polygons and enriched with metadata. This methodology builds a qualitative annotated training corpus of approximately 20,000 cropped signs (i.e. 240,000 visualizations), all with their unicode codepoint and conventional sign name. It will act as a multi-layerd core dataset for the further development and fine-tuning of a machine learning OCR training model for the Old Babylonian cuneiform script. This paper discusses how the physical nature of handwritten inscribed Old Babylonian documentary clay tablets challenges the annotation and metadating task, and how these have been addressed within the CUNE-IIIF-ORM project to achieve an effective training corpus to support the training of a machine learning OCR model. ACM CCS Applied computing → Document management and text processing → Document capture → Optical character recognition; Applied computing → Arts and humanities → Language translation.

Algorithm for extracting contours of agricultural crops images

Abstract. Currently, identification of crop diseases and their prevention is one of the main problems in the field of agriculture. Conventional visual inspection is a time and money consuming process for farms. Therefore, images taken by unmanned aerial devices or satellites are used to assess the condition of crops, control them and identify diseases. In particular, when identifying crop diseases, it is necessary to first solve the problem of automatic recognition of their type through the image of crops. Usually contour separation algorithms are widely used in the segmentation of objects in the image. This work is aimed at solving the problem of separating the contour of the object, in which algorithms are formed based on Canny, Sobel and Robinson filters, which are considered to be popular and classical methods of contour separation, and their various combinations. In the computational experiments, a set of contour images, whose contours were separated by an expert, was used. Evaluation was performed by comparing the image obtained by applying the combination of filters to the original image and the corresponding contour image pixels separated by an expert. The proposed approach has been tested on a set of plant leaf images and shown to be effective.