Ginger Seeding Detection and Shoot Orientation Discrimination Using an Improved YOLOv4-LITE Network

A consistent orientation of ginger shoots when sowing ginger is more conducive to high yields and later harvesting. However, current ginger sowing mainly relies on manual methods, seriously hindering the ginger industry’s development. Existing ginger seeders still require manual assistance in placing ginger seeds to achieve consistent ginger shoot orientation. To address the problem that existing ginger seeders have difficulty in automating seeding and ensuring consistent ginger shoot orientation, this study applies object detection techniques in deep learning to the detection of ginger and proposes a ginger recognition network based on YOLOv4-LITE, which, first, uses MobileNetv2 as the backbone network of the model and, second, adds coordinate attention to MobileNetv2 and uses Do-Conv convolution to replace part of the traditional convolution. After completing the prediction of ginger and ginger shoots, this paper determines ginger shoot orientation by calculating the relative positions of the largest ginger shoot and the ginger. The mean average precision, Params, and giga Flops of the proposed YOLOv4-LITE in the test set reached 98.73%, 47.99 M, and 8.74, respectively. The experimental results show that YOLOv4-LITE achieved ginger seed detection and ginger shoot orientation calculation, and that it provides a technical guarantee for automated ginger seeding.

Download Full-text

Deep learning-based detection of dental prostheses and restorations

Scientific Reports ◽

10.1038/s41598-021-81202-x ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Toshihito Takahashi ◽

Kazunori Nozaki ◽

Tomoya Gonda ◽

Tomoaki Mameno ◽

Kazunori Ikebe

Keyword(s):

Deep Learning ◽

High Accuracy ◽

Mean Average Precision ◽

Learning Performance ◽

Average Precision ◽

Tooth Color ◽

Photographic Images ◽

Dental Prostheses ◽

Learning Procedure ◽

The Mean

AbstractThe purpose of this study is to develop a method for recognizing dental prostheses and restorations of teeth using a deep learning. A dataset of 1904 oral photographic images of dental arches (maxilla: 1084 images; mandible: 820 images) was used in the study. A deep-learning method to recognize the 11 types of dental prostheses and restorations was developed using TensorFlow and Keras deep learning libraries. After completion of the learning procedure, the average precision of each prosthesis, mean average precision, and mean intersection over union were used to evaluate learning performance. The average precision of each prosthesis varies from 0.59 to 0.93. The mean average precision and mean intersection over union of this system were 0.80 and 0.76, respectively. More than 80% of metallic dental prostheses were detected correctly, but only 60% of tooth-colored prostheses were detected. The results of this study suggest that dental prostheses and restorations that are metallic in color can be recognized and predicted with high accuracy using deep learning; however, those with tooth color are recognized with moderate accuracy.

Download Full-text

Deep Learning for Plant Classification and Content-Based Image Retrieval

Cybernetics and Information Technologies ◽

10.2478/cait-2019-0005 ◽

2019 ◽

Vol 19 (1) ◽

pp. 88-100

Author(s):

Bálint Pál Gyires-Tóth ◽

Márton Osváth ◽

Dávid Papp ◽

Gábor Szűcs

Keyword(s):

Neural Network ◽

Deep Learning ◽

Image Retrieval ◽

Search Task ◽

Feature Learning ◽

Content Based Image Retrieval ◽

Network Architectures ◽

Average Precision ◽

Test Set ◽

Fully Connected

Abstract The main goal of the present research is to classify images of plants to species with deep learning. We used convolutional neural network architectures for feature learning and fully connected layers with logsoftmax output for classification. Pretrained models on ImageNet were used, and transfer learning was applied. In the current research image sets published in the scope of the PlantCLEF 2015 challenge were used. The proposed system surpasses the results of all top competitors of the challenge by 8% and 7% at observation and image levels, respectively. Our secondary goal was to satisfy the users’ needs in content-based image retrieval to give relevant hits during species search task. We optimized the length of the returned lists in order to maximize MAP (Mean Average Precision), which is critical to the performance of image retrieval. Thus, we achieved more than 50% improvement of MAP in the test set compared to the baseline.

Download Full-text

DEFORESTATION DETECTION IN THE AMAZON RAINFOREST WITH SPATIAL AND CHANNEL ATTENTION MECHANISMS

ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-archives-xliii-b3-2021-851-2021 ◽

2021 ◽

Vol XLIII-B3-2021 ◽

pp. 851-858

Author(s):

P. Tovar ◽

M. O. Adarme ◽

R. Q. Feitosa

Keyword(s):

Neural Networks ◽

Deep Learning ◽

Loss Function ◽

Satellite Images ◽

Attention Mechanism ◽

Amazon Rainforest ◽

Average Precision ◽

Manual Task ◽

Optical Images ◽

The Mean

Abstract. Deforestation in the Amazon rainforest is an alarming problem of global interest. Environmental impacts of this process are countless, but probably the most significant concerns regard the increase in CO2 emissions and global temperature rise. Currently, the assessment of deforested areas in the Amazon region is a manual task, where people analyse multiple satellite images to quantify the deforestation. We propose a method for automatic deforestation detection based on Deep Learning Neural Networks with dual-attention mechanisms. We employed a siamese architecture to detect deforestation changes between optical images in 2018 and 2019. Experiments were performed to evaluate the relevance and sensitivity of hyperparameter tuning of the loss function and the effects of dual-attention mechanisms (spatial and channel) in predicting deforestation. Experimental results suggest that a proper tuning of the loss function might bring benefits in terms of generalisation. We also show that the spatial attention mechanism is more relevant for deforestation detection than the channel attention mechanism. When both mechanisms are combined, the greatest improvements are found, and we reported an increase of 1.06% in the mean average precision over a baseline.

Download Full-text

Improving Document Retrieval with Spelling Correction for Weak and Fabricated Indonesian-Translated Hadith

Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) ◽

10.29207/resti.v4i3.1913 ◽

2020 ◽

Vol 4 (3) ◽

pp. 551-557

Author(s):

Muhammad zaky ramadhan ◽

Kemas Muslim Lhaksmana

Keyword(s):

Retrieval System ◽

Islamic Law ◽

Vector Space Model ◽

Document Retrieval ◽

The Internet ◽

Average Precision ◽

Spelling Correction ◽

Space Model ◽

The Mean ◽

The Web

Hadith has several levels of authenticity, among which are weak (dhaif), and fabricated (maudhu) hadith that may not originate from the prophet Muhammad PBUH, and thus should not be considered in concluding an Islamic law (sharia). However, many such hadiths have been commonly confused as authentic hadiths among ordinary Muslims. To easily distinguish such hadiths, this paper proposes a method to check the authenticity of a hadith by comparing them with a collection of fabricated hadiths in Indonesian. The proposed method applies the vector space model and also performs spelling correction using symspell to check whether the use of spelling check can improve the accuracy of hadith retrieval, because it has never been done in previous works and typos are common on Indonesian-translated hadiths on the Web and social media raw text. The experiment result shows that the use of spell checking improves the mean average precision and recall to become 81% (from 73%) and 89% (from 80%), respectively. Therefore, the improvement in accuracy by implementing spelling correction make the hadith retrieval system more feasible and encouraged to be implemented in future works because it can correct typos that are common in the raw text on the Internet.

Download Full-text

Pencarian Question-Answer Menggunakan Convolutional Neural Network Pada Topik Agama Berbahasa Indonesia

Jurnal ULTIMATICS ◽

10.31937/ti.v10i1.842 ◽

2018 ◽

Vol 10 (1) ◽

pp. 57-64 ◽

Cited By ~ 1

Author(s):

Rizqa Raaiqa Bintana ◽

Chastine Fatichah ◽

Diana Purwitasari

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Question Answering ◽

Mean Average Precision ◽

Average Precision ◽

Community Based ◽

Network Methods ◽

The Mean ◽

Index Terms ◽

Search Information

Community-based question answering (CQA) is formed to help people who search information that they need through a community. One condition that may occurs in CQA is when people cannot obtain the information that they need, thus they will post a new question. This condition can cause CQA archive increased because of duplicated questions. Therefore, it becomes important problems to find semantically similar questions from CQA archive towards a new question. In this study, we use convolutional neural network methods for semantic modeling of sentence to obtain words that they represent the content of documents and new question. The result for the process of finding the same question semantically to a new question (query) from the question-answer documents archive using the convolutional neural network method, obtained the mean average precision value is 0,422. Whereas by using vector space model, as a comparison, obtained mean average precision value is 0,282. Index Terms—community-based question answering, convolutional neural network, question retrieval

Download Full-text

Android Malware Detection Techniques: A Literature Review

Recent Patents on Engineering ◽

10.2174/1872212114999200710143847 ◽

2020 ◽

Vol 14 ◽

Author(s):

Meghna Dhalaria ◽

Ekta Gandotra

Keyword(s):

Machine Learning ◽

Deep Learning ◽

Malware Detection ◽

Future Research ◽

Android Malware ◽

Detection Techniques ◽

Android Malware Detection ◽

Future Research Directions ◽

To Come ◽

Tools And Techniques

Purpose: This paper provides the basics of Android malware, its evolution and tools and techniques for malware analysis. Its main aim is to present a review of the literature on Android malware detection using machine learning and deep learning and identify the research gaps. It provides the insights obtained through literature and future research directions which could help researchers to come up with robust and accurate techniques for classification of Android malware. Design/Methodology/Approach: This paper provides a review of the basics of Android malware, its evolution timeline and detection techniques. It includes the tools and techniques for analyzing the Android malware statically and dynamically for extracting features and finally classifying these using machine learning and deep learning algorithms. Findings: The number of Android users is expanding very fast due to the popularity of Android devices. As a result, there are more risks to Android users due to the exponential growth of Android malware. On-going research aims to overcome the constraints of earlier approaches for malware detection. As the evolving malware are complex and sophisticated, earlier approaches like signature based and machine learning based are not able to identify these timely and accurately. The findings from the review shows various limitations of earlier techniques i.e. requires more detection time, high false positive and false negative rate, low accuracy in detecting sophisticated malware and less flexible. Originality/value: This paper provides a systematic and comprehensive review on the tools and techniques being employed for analysis, classification and identification of Android malicious applications. It includes the timeline of Android malware evolution, tools and techniques for analyzing these statically and dynamically for the purpose of extracting features and finally using these features for their detection and classification using machine learning and deep learning algorithms. On the basis of the detailed literature review, various research gaps are listed. The paper also provides future research directions and insights which could help researchers to come up with innovative and robust techniques for detecting and classifying the Android malware.

Download Full-text

Cotton Stand Counting from Unmanned Aerial System Imagery Using MobileNet and CenterNet Deep Learning Models

Remote Sensing ◽

10.3390/rs13142822 ◽

2021 ◽

Vol 13 (14) ◽

pp. 2822

Author(s):

Zhe Lin ◽

Wenxuan Guo

Keyword(s):

Deep Learning ◽

Cotton Plant ◽

Unmanned Aerial System ◽

Learning Models ◽

Training Images ◽

Testing Dataset ◽

Cotton Plants ◽

Detection And Counting ◽

Different Dimensions ◽

The Mean

An accurate stand count is a prerequisite to determining the emergence rate, assessing seedling vigor, and facilitating site-specific management for optimal crop production. Traditional manual counting methods in stand assessment are labor intensive and time consuming for large-scale breeding programs or production field operations. This study aimed to apply two deep learning models, the MobileNet and CenterNet, to detect and count cotton plants at the seedling stage with unmanned aerial system (UAS) images. These models were trained with two datasets containing 400 and 900 images with variations in plant size and soil background brightness. The performance of these models was assessed with two testing datasets of different dimensions, testing dataset 1 with 300 by 400 pixels and testing dataset 2 with 250 by 1200 pixels. The model validation results showed that the mean average precision (mAP) and average recall (AR) were 79% and 73% for the CenterNet model, and 86% and 72% for the MobileNet model with 900 training images. The accuracy of cotton plant detection and counting was higher with testing dataset 1 for both CenterNet and MobileNet models. The results showed that the CenterNet model had a better overall performance for cotton plant detection and counting with 900 training images. The results also indicated that more training images are required when applying object detection models on images with different dimensions from training datasets. The mean absolute percentage error (MAPE), coefficient of determination (R2), and the root mean squared error (RMSE) values of the cotton plant counting were 0.07%, 0.98 and 0.37, respectively, with testing dataset 1 for the CenterNet model with 900 training images. Both MobileNet and CenterNet models have the potential to accurately and timely detect and count cotton plants based on high-resolution UAS images at the seedling stage. This study provides valuable information for selecting the right deep learning tools and the appropriate number of training images for object detection projects in agricultural applications.

Download Full-text

Design and Implementation of Intelligent Inspection and Alarm Flight System for Epidemic Prevention

Drones ◽

10.3390/drones5030068 ◽

2021 ◽

Vol 5 (3) ◽

pp. 68

Author(s):

Jiwei Fan ◽

Xiaogang Yang ◽

Ruitao Lu ◽

Xueli Xie ◽

Weipeng Li

Keyword(s):

Deep Learning ◽

Autonomous Navigation ◽

Detection Method ◽

Active Role ◽

Absolute Error ◽

Face Mask ◽

Learning Technology ◽

Flight System ◽

Crowd Density ◽

The Mean

Unmanned aerial vehicles (UAV) and related technologies have played an active role in the prevention and control of novel coronaviruses at home and abroad, especially in epidemic prevention, surveillance, and elimination. However, the existing UAVs have a single function, limited processing capacity, and poor interaction. To overcome these shortcomings, we designed an intelligent anti-epidemic patrol detection and warning flight system, which integrates UAV autonomous navigation, deep learning, intelligent voice, and other technologies. Based on the convolution neural network and deep learning technology, the system possesses a crowd density detection method and a face mask detection method, which can detect the position of dense crowds. Intelligent voice alarm technology was used to achieve an intelligent alarm system for abnormal situations, such as crowd-gathering areas and people without masks, and to carry out intelligent dissemination of epidemic prevention policies, which provides a powerful technical means for epidemic prevention and delaying their spread. To verify the superiority and feasibility of the system, high-precision online analysis was carried out for the crowd in the inspection area, and pedestrians’ faces were detected on the ground to identify whether they were wearing a mask. The experimental results show that the mean absolute error (MAE) of the crowd density detection was less than 8.4, and the mean average precision (mAP) of face mask detection was 61.42%. The system can provide convenient and accurate evaluation information for decision-makers and meets the requirements of real-time and accurate detection.

Download Full-text

Enhancement of Detecting Permanent Water and Temporary Water in Flood Disasters by Fusing Sentinel-1 and Sentinel-2 Imagery Using Deep Learning Algorithms: Demonstration of Sen1Floods11 Benchmark Datasets

Remote Sensing ◽

10.3390/rs13112220 ◽

2021 ◽

Vol 13 (11) ◽

pp. 2220

Author(s):

Yanbing Bai ◽

Wenqi Wu ◽

Zhengxin Yang ◽

Jinze Yu ◽

Bo Zhao ◽

...

Keyword(s):

Remote Sensing ◽

Deep Learning ◽

Large Scale ◽

Learning Algorithms ◽

Flood Events ◽

Test Set ◽

Source Data ◽

Flood Disasters ◽

Temporary Water ◽

Sentinel 2

Identifying permanent water and temporary water in flood disasters efficiently has mainly relied on change detection method from multi-temporal remote sensing imageries, but estimating the water type in flood disaster events from only post-flood remote sensing imageries still remains challenging. Research progress in recent years has demonstrated the excellent potential of multi-source data fusion and deep learning algorithms in improving flood detection, while this field has only been studied initially due to the lack of large-scale labelled remote sensing images of flood events. Here, we present new deep learning algorithms and a multi-source data fusion driven flood inundation mapping approach by leveraging a large-scale publicly available Sen1Flood11 dataset consisting of roughly 4831 labelled Sentinel-1 SAR and Sentinel-2 optical imagery gathered from flood events worldwide in recent years. Specifically, we proposed an automatic segmentation method for surface water, permanent water, and temporary water identification, and all tasks share the same convolutional neural network architecture. We utilize focal loss to deal with the class (water/non-water) imbalance problem. Thorough ablation experiments and analysis confirmed the effectiveness of various proposed designs. In comparison experiments, the method proposed in this paper is superior to other classical models. Our model achieves a mean Intersection over Union (mIoU) of 52.99%, Intersection over Union (IoU) of 52.30%, and Overall Accuracy (OA) of 92.81% on the Sen1Flood11 test set. On the Sen1Flood11 Bolivia test set, our model also achieves very high mIoU (47.88%), IoU (76.74%), and OA (95.59%) and shows good generalization ability.

Download Full-text

Image Enhanced Mask R-CNN: A Deep Learning Pipeline with New Evaluation Measures for Wind Turbine Blade Defect Detection and Classification

Journal of Imaging ◽

10.3390/jimaging7030046 ◽

2021 ◽

Vol 7 (3) ◽

pp. 46

Author(s):

Jiajun Zhang ◽

Georgina Cosma ◽

Jason Watkins

Keyword(s):

Deep Learning ◽

Wind Turbine ◽

Turbine Blade ◽

Defect Detection ◽

Recognition Rate ◽

Weighted Average ◽

Wind Turbine Blade ◽

Evaluation Measures ◽

Augmentation Techniques ◽

The Mean

Demand for wind power has grown, and this has increased wind turbine blade (WTB) inspections and defect repairs. This paper empirically investigates the performance of state-of-the-art deep learning algorithms, namely, YOLOv3, YOLOv4, and Mask R-CNN for detecting and classifying defects by type. The paper proposes new performance evaluation measures suitable for defect detection tasks, and these are: Prediction Box Accuracy, Recognition Rate, and False Label Rate. Experiments were carried out using a dataset, provided by the industrial partner, that contains images from WTB inspections. Three variations of the dataset were constructed using different image augmentation settings. Results of the experiments revealed that on average, across all proposed evaluation measures, Mask R-CNN outperformed all other algorithms when transformation-based augmentations (i.e., rotation and flipping) were applied. In particular, when using the best dataset, the mean Weighted Average (mWA) values (i.e., mWA is the average of the proposed measures) achieved were: Mask R-CNN: 86.74%, YOLOv3: 70.08%, and YOLOv4: 78.28%. The paper also proposes a new defect detection pipeline, called Image Enhanced Mask R-CNN (IE Mask R-CNN), that includes the best combination of image enhancement and augmentation techniques for pre-processing the dataset, and a Mask R-CNN model tuned for the task of WTB defect detection and classification.

Download Full-text