Advances and Limitations in Open Source Arabic-Script OCR: A Case Study

This work presents an accuracy study of the open source OCR engine, Kraken, on the leading Arabic scholarly journal, al-Abhath. In contrast with other commercially available OCR engines, Kraken is shown to be capable of producing highly accurate Arabic-script OCR. The study also assesses the relative accuracy of typeface-specific and generalized models on the al-Abhath data and provides a microanalysis of the “error instances” and the contextual features that may have contributed to OCR misrecognition. Building on this analysis, the paper argues that Arabic-script OCR can be significantly improved through (1) a more systematic approach to training data production, and (2) the development of key technological components, especially multi-language models and improved line segmentation and layout analysis./Cet article présente une étude d’exactitude du moteur ROC open source, Krakan, sur la revue académique arabe de premier rang, al-Abhath. Contrairement à d’autres moteurs ROC disponibles sur le marché, Kraken se révèle être capable de produire de la ROC extrêmement exacte de l’écriture arabe. L’étude évalue aussi l’exactitude relative des modèles spécifiquement configurés à des polices et celle des modèles généralisés sur les données d’al-Abhath et fournit une microanalyse des « occurrences d’erreurs », ainsi qu’une microanalyse des éléments contextuels qui pourraient avoir contribué à la méreconnaissance ROC. S’appuyant sur cette analyse, cet article fait valoir que la ROC de l’écriture arabe peut être considérablement améliorée grâce à (1) une approche plus systématique d’entraînement de la production de données et (2) grâce au développement de composants technologiques fondamentaux, notammentl’amélioration des modèles multilingues, de la segmentation de ligne et de l’analyse de la mise en page.

Download Full-text

A Case Study on the Development of Life Insurance Prediction Model Analysis Environment Applying Open-Source Bigdata Technology

Productivity Review ◽

10.15843/kpapr.32.1.2018.03.173 ◽

2018 ◽

Vol 32 (1) ◽

pp. 173-207

Author(s):

Sung Hyun Kim ◽

◽

Jiwon Chun ◽

Miyoung Lee

Keyword(s):

Prediction Model ◽

Open Source ◽

Life Insurance ◽

Model Analysis ◽

Analysis Environment

Download Full-text

Performance Comparison of Web Backend And Database: A Case Study Of Node.JS, Golang and MySQL, MongoDB

Recent Advances in Computer Science and Communications ◽

10.2174/2666255813666191219104133 ◽

2019 ◽

Vol 13 ◽

Author(s):

Faried Effendy ◽

Taufik ◽

Bramantyo Adhilaksono

Keyword(s):

Response Time ◽

Open Source ◽

Poisson Distribution ◽

Mobile Application ◽

Web Applications ◽

Performance Comparison ◽

Memory Usage ◽

Data Traffic ◽

Cpu Utilization

: Substantial research has been conducted to compare web servers or to compare databases, but very limited research combines the two. Node.js and Golang (Go) are popular platforms for both web and mobile application back-ends, whereas MySQL and Go are among the best open source databases with different characters. Using MySQL and MongoDB as databases, this study aims to compare the performance of Go and Node.js as web applications back-end regarding response time, CPU utilization, and memory usage. To simulate the actual web server workload, the flow of data traffic on the server follows the Poisson distribution. The result shows that the combination of Go and MySQL is superior in CPU utilization and memory usage, while the Node.js and MySQL combination is superior in response time.

Download Full-text

DEVELOPING A FREE AND OPEN-SOURCE GRAPHICAL USER INTERFACE FOR CNN-BASED GEOMORPHIC FEATURE DETECTION: CAROLINA BAYS AS A CASE STUDY

10.1130/abs/2020am-354782 ◽

2020 ◽

Author(s):

Mark Lundine ◽

◽

Arthur Trembanis

Keyword(s):

User Interface ◽

Open Source ◽

Graphical User Interface ◽

Feature Detection ◽

Carolina Bays ◽

Geomorphic Feature

Download Full-text

Differences in echocardiography interpretation techniques among trainees and expert readers

Journal of Echocardiography ◽

10.1007/s12574-021-00531-y ◽

2021 ◽

Author(s):

David Roy Anderson ◽

Sarah Blissett ◽

Patricia O’Sullivan ◽

Atif Qasim

Keyword(s):

Diagnostic Accuracy ◽

Systematic Approach ◽

Qualitative Case Study ◽

First Year ◽

Educational Tool ◽

Thought Processes ◽

Advanced Imaging ◽

Multiple Images ◽

Novice Readers

Abstract Background Trainees learn transthoracic echocardiogram (TTE) interpretation through independently completing and reviewing selected portions of the study with experts. The diagnostic accuracy of novice TTE interpretation is known to be low and schema for reading TTEs systematically are lacking. The purpose of our study is to identify techniques experts use while reading TTEs which could be used to more effectively teach novice readers. Methods We performed a prospective qualitative case study to observe how experts and trainees interpret TTEs in an academic institution using a concurrent think aloud (CTA) method. Three TTEs of intermediate complexity were given to 3 advanced imaging fellows, 3 first year fellows and 3 expert TTE readers Participants filled out a report while reading and described aloud their thought processes. Sessions were video and audiotaped for analysis. Results Experts and advanced fellows used specific techniques that novices did not including: previewing studies, reviewing multiple images simultaneously, having flexibility in image review order and disease coding, and saving hardest elements to code for the end. Direct observation of TTE reading informed trainee inefficiencies and was a well-received educational tool. Conclusions In this single centered study we identified several unique approaches experts use to interpret TTEs which may be teachable to novices. Although limited in generalizability the findings of this study suggests that a more systematic approach to TTE interpretation, using techniques found in experts, might be of significant value for trainees. Further study is needed to evaluate teaching practices at other institutions and to assess whether implementation of these techniques by novices improves can improve their diagnostic accuracy and efficiency of reading at an earlier stage in their training.

Download Full-text

Vehicle Energy Consumption in Python (VencoPy): Presenting and Demonstrating an Open-Source Tool to Calculate Electric Vehicle Charging Flexibility

Energies ◽

10.3390/en14144349 ◽

2021 ◽

Vol 14 (14) ◽

pp. 4349

Author(s):

Niklas Wulff ◽

Fabia Miorelli ◽

Hans Christian Gils ◽

Patrick Jochem

Keyword(s):

Open Source ◽

Electric Vehicle ◽

Full Range ◽

Mobility Patterns ◽

Electric Vehicle Charging ◽

Survey Results ◽

Vehicle Demand ◽

Vehicle Fleet ◽

Electric Loads

As electric vehicle fleets grow, rising electric loads necessitate energy systems models to incorporate their respective demand and potential flexibility. Recently, a small number of tools for electric vehicle demand and flexibility modeling have been released under open source licenses. These usually sample discrete trips based on aggregate mobility statistics. However, the full range of variables of travel surveys cannot be accessed in this way and sub-national mobility patterns cannot be modeled. Therefore, a tool is proposed to estimate future electric vehicle fleet charging flexibility while being able to directly access detailed survey results. The framework is applied in a case study involving two recent German national travel surveys (from the years 2008 and 2017) to exemplify the implications of different mobility patterns of motorized individual vehicles on load shifting potential of electric vehicle fleets. The results show that different mobility patterns, have a significant impact on the resulting load flexibilites. Most obviously, an increased daily mileage results in higher electricty demand. A reduced number of trips per day, on the other hand, leads to correspondingly higher grid connectivity of the vehicle fleet. VencoPy is an open source, well-documented and maintained tool, capable of assessing electric vehicle fleet scenarios based on national travel surveys. To scrutinize the tool, a validation of the simulated charging by empirically observed electric vehicle fleet charging is advised.

Download Full-text

Investigating the Potential of Network Optimization for a Constrained Object Detection Problem

Journal of Imaging ◽

10.3390/jimaging7040064 ◽

2021 ◽

Vol 7 (4) ◽

pp. 64

Author(s):

Tanguy Ophoff ◽

Cédric Gullentops ◽

Kristof Van Beeck ◽

Toon Goedemé

Keyword(s):

Computational Complexity ◽

Object Detection ◽

Network Optimization ◽

Real Life ◽

Optimization Techniques ◽

Training Data ◽

Single Shot ◽

Standard Object ◽

Number Of Classes

Object detection models are usually trained and evaluated on highly complicated, challenging academic datasets, which results in deep networks requiring lots of computations. However, a lot of operational use-cases consist of more constrained situations: they have a limited number of classes to be detected, less intra-class variance, less lighting and background variance, constrained or even fixed camera viewpoints, etc. In these cases, we hypothesize that smaller networks could be used without deteriorating the accuracy. However, there are multiple reasons why this does not happen in practice. Firstly, overparameterized networks tend to learn better, and secondly, transfer learning is usually used to reduce the necessary amount of training data. In this paper, we investigate how much we can reduce the computational complexity of a standard object detection network in such constrained object detection problems. As a case study, we focus on a well-known single-shot object detector, YoloV2, and combine three different techniques to reduce the computational complexity of the model without reducing its accuracy on our target dataset. To investigate the influence of the problem complexity, we compare two datasets: a prototypical academic (Pascal VOC) and a real-life operational (LWIR person detection) dataset. The three optimization steps we exploited are: swapping all the convolutions for depth-wise separable convolutions, perform pruning and use weight quantization. The results of our case study indeed substantiate our hypothesis that the more constrained a problem is, the more the network can be optimized. On the constrained operational dataset, combining these optimization techniques allowed us to reduce the computational complexity with a factor of 349, as compared to only a factor 9.8 on the academic dataset. When running a benchmark on an Nvidia Jetson AGX Xavier, our fastest model runs more than 15 times faster than the original YoloV2 model, whilst increasing the accuracy by 5% Average Precision (AP).

Download Full-text

PVDP: A portable open source pipeline for detection of plant viruses in RNAseq data. A case study on potato viruses in Antioquia (Colombia)

Physiological and Molecular Plant Pathology ◽

10.1016/j.pmpp.2021.101604 ◽

2021 ◽

Vol 113 ◽

pp. 101604

Author(s):

Pablo Gutiérrez ◽

Ary Rivillas ◽

Daniel Tejada ◽

Susana Giraldo ◽

Andrea Restrepo ◽

...

Keyword(s):

Open Source ◽

Plant Viruses ◽

Potato Viruses ◽

Rnaseq Data

Download Full-text

Augmenting Crop Detection for Precision Agriculture with Deep Visual Transfer Learning—A Case Study of Bale Detection

Remote Sensing ◽

10.3390/rs13010023 ◽

2020 ◽

Vol 13 (1) ◽

pp. 23

Author(s):

Wei Zhao ◽

William Yamada ◽

Tianxin Li ◽

Matthew Digman ◽

Troy Runge

Keyword(s):

Object Detection ◽

Transfer Learning ◽

Precision Agriculture ◽

Crop Production ◽

Domain Adaptation ◽

Training Data ◽

Detection Accuracy ◽

Detection Model ◽

Agriculture Products

In recent years, precision agriculture has been researched to increase crop production with less inputs, as a promising means to meet the growing demand of agriculture products. Computer vision-based crop detection with unmanned aerial vehicle (UAV)-acquired images is a critical tool for precision agriculture. However, object detection using deep learning algorithms rely on a significant amount of manually prelabeled training datasets as ground truths. Field object detection, such as bales, is especially difficult because of (1) long-period image acquisitions under different illumination conditions and seasons; (2) limited existing prelabeled data; and (3) few pretrained models and research as references. This work increases the bale detection accuracy based on limited data collection and labeling, by building an innovative algorithms pipeline. First, an object detection model is trained using 243 images captured with good illimitation conditions in fall from the crop lands. In addition, domain adaptation (DA), a kind of transfer learning, is applied for synthesizing the training data under diverse environmental conditions with automatic labels. Finally, the object detection model is optimized with the synthesized datasets. The case study shows the proposed method improves the bale detecting performance, including the recall, mean average precision (mAP), and F measure (F1 score), from averages of 0.59, 0.7, and 0.7 (the object detection) to averages of 0.93, 0.94, and 0.89 (the object detection + DA), respectively. This approach could be easily scaled to many other crop field objects and will significantly contribute to precision agriculture.

Download Full-text