Attention-Enhanced YOLO Model via Region of Interest Feature Extraction for Leaf Diseases Detection

2021 ◽  
Vol 19 (4) ◽  
pp. 83-93
Author(s):  
Tae-Min Choi ◽  
Chang-Hwan Son ◽  
Donghyuk Lee
Keyword(s):  
Feature Extraction ◽  
Region Of Interest

Feature Extraction by Decryption to Target

2014 ◽  
Vol 1008-1009 ◽  
pp. 1509-1512
Author(s):  
Qing E Wu ◽  
Hong Wang ◽  
Li Fen Ding
Keyword(s):  
Feature Extraction ◽  
Wavelet Transform ◽  
Extraction Method ◽  
Recognition Rate ◽  
Region Of Interest ◽  
Correct Recognition ◽  
Palmprint Recognition ◽  
Feature Extraction Method ◽  
Energy Feature ◽  
Correct Recognition Rate

To carry out an effective classification and recognition for target, this paper studied the target owned characteristics, discussed a decryption algorithm, gave a feature extraction method based on the decryption process, and extracted the feature of palmprint in region of interest. Moreover, this paper used the wavelet transform to extract the energy feature of target, gave an approach on matching and recognition to improve the correctness and efficiency of existing recognition approaches, and compared it with existing approaches of palmprint recognition by experiments. The experiment results show that the correct recognition rate of the approach in this paper is improved averagely by 2.34% than that of the existing recognition approaches.

scholarly journals Perbandingan Metode Klasifikasi pada Pengolahan Citra Mata Ikan Tuna

2020 ◽  
Vol 5 ◽  
Author(s):  
Toni Dwi Novianto ◽  
I Made Susi Erawan
Keyword(s):  
Image Processing ◽  
Feature Extraction ◽  
Region Of Interest ◽  
Nearest Neighbors ◽  
Support Vector ◽  
Gray Level ◽  
K Nearest Neighbors ◽  
Eye Color ◽  
Occurrence Matrix ◽  
Fish Eye

<p class="AbstractEnglish"><strong>Abstract:</strong> Fish eye color is an important attribute of fish quality. The change in eye color during the storage process correlates with freshness and has a direct effect on consumer perception. The process of changing the color of the fish eye can be analyzed using image processing. The purpose of this study was to obtain the best classification method for predicting fish freshness based on image processing in fish eyes. Three tuna fish were used in this study. The test was carried out for 20 hours with an eye image every 2 hours at room temperature. Fish eye image processing uses Matlab R.2017a software while the classification uses Weka 3.8 software. The image processing stages are taking fish eye image, segmenting ROI (region of interest), converting RGB image to grayscale, and feature extraction. Feature extraction used is the gray-level co-occurrence matrix (GLCM). The classification techniques used are artificial neural networks (ANN), k-neighborhood neighbors (k-NN), and support vector machines (SVM). The results showed the value using ANN = 0.53, k-NN = 0.83, and SVM = 0.69. Based on these results it can be determined that the best classification technique is to use the k-nearest neighbor (k-NN).</p><p class="AbstrakIndonesia"><strong>Abstrak:</strong> Warna mata ikan merupakan atribut penting pada kualitas ikan. Perubahan warna mata ikan selama proses penyimpanan berhubungan dengan tingkat kesegaran dan memiliki efek langsung pada persepsi konsumen. Proses perubahan warna mata ikan dapat dianalisis menggunakan pengolahan citra. Tujuan penelitian ini adalah mendapatkan metode klasifikasi terbaik untuk memprediksi kesegaran ikan berbasis pengolahan citra pada mata ikan. Tiga ekor ikan tuna digunakan dalam penelitian ini. Pengujian dilakukan selama 20 jam dengan pengambilan citra mata setiap 2 jam pada suhu ruang. Pengolahan citra mata ikan menggunakan software matlab R.2017a sedangkan pengklasifiannya menggunakan software Weka 3.8. Tahapan pengolahan citra meliputi pengambilan citra mata ikan, segmentasi ROI (<em>region of interest</em>), konversi citra RGB menjadi <em>grayscale</em>, dan ekstraksi fitur. Ekstraksi fitur yang digunakan yaitu <em>gray-level co-occurrence matrix</em> (GLCM).  Teknik klasifikasi yang digunakan yaitu, <em>artificial neural network</em> (ANN), <em>k-nearest neighbors</em> (k-NN), dan <em>support vector machine</em> (SVM). Hasil penelitian menunjukkan nilai korelasi menggunakan ANN = 0,53, k-NN = 0,83, dan SVM = 0,69. Berdasarkan hasil tersebut dapat disimpulkan teknik klasifikasi terbaik adalah menggunakan <em>k-nearest neighbors</em> (k-NN).</p>

scholarly journals Apple Leaf Disease Identification Through Region-of-Interest-Aware Deep Convolutional Neural Network

2020 ◽  
Vol 64 (2) ◽  
pp. 20507-1-20507-10 ◽  
Author(s):  
Hee-Jin Yu ◽  
Chang-Hwan Son ◽  
Dong Hyuk Lee
Keyword(s):  
Neural Network ◽  
Feature Extraction ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Leaf Area ◽  
Region Of Interest ◽  
Deep Convolutional Neural Network ◽  
Discriminative Power ◽  
Feature Map ◽  
Traditional Approaches

Abstract Traditional approaches for the identification of leaf diseases involve the use of handcrafted features such as colors and textures for feature extraction. Therefore, these approaches may have limitations in extracting abundant and discriminative features. Although deep learning approaches have been recently introduced to overcome the shortcomings of traditional approaches, existing deep learning models such as VGG and ResNet have been used in these approaches. This indicates that the approach can be further improved to increase the discriminative power because the spatial attention mechanism to predict the background and spot areas (i.e., local areas with leaf diseases) has not been considered. Therefore, a new deep learning architecture, which is hereafter referred to as region-of-interest-aware deep convolutional neural network (ROI-aware DCNN), is proposed to make deep features more discriminative and increase classification performance. The primary idea is that leaf disease symptoms appear in leaf area, whereas the background region does not contain useful information regarding leaf diseases. To realize this, two subnetworks are designed. One subnetwork is the ROI subnetwork to provide more discriminative features from the background, leaf areas, and spot areas in the feature map. The other subnetwork is the classification subnetwork to increase the classification accuracy. To train the ROI-aware DCNN, the ROI subnetwork is first learned with a new image set containing the ground truth images where the background, leaf area, and spot area are divided. Subsequently, the entire network is trained in an end-to-end manner to connect the ROI subnetwork with the classification subnetwork through a concatenation layer. The experimental results confirm that the proposed ROI-aware DCNN can increase the discriminative power by predicting the areas in the feature map that are more important for leaf diseases identification. The results prove that the proposed method surpasses conventional state-of-the-art methods such as VGG, ResNet, SqueezeNet, bilinear model, and multiscale-based deep feature extraction and pooling.

Smart Surveillance System Detection Using PCA- Based Feature Extraction

2012 ◽  
Vol 232 ◽  
pp. 137-141 ◽  
Author(s):  
Ahmad Kadri Junoh ◽  
Muhammad Naufal Mansor ◽  
Mohd Shafarudy Abu ◽  
Wan Zuki Azman Wan Ahmad ◽  
Wan Nur Hadani Wan Jaafar ◽  
...  
Keyword(s):  
Feature Extraction ◽  
Face Recognition ◽  
Surveillance System ◽  
Region Of Interest ◽  
Expression Recognition ◽  
Interdisciplinary Field ◽  
Human Activity Analysis ◽  
Actual Mechanism ◽  
Smart Surveillance ◽  
Robust Recognition

Perception of vision and motion is a vast interdisciplinary field combining psychology, neurology, physiology, mathematics, computer science, physics, philosophy and more. The issue of the actual mechanism for the visual and computational perception of motion in the human are keep grow for the last decade. Each of the researchers is keep pursuit to find the ideal potion of a robust recognition and detection for video system. Clutches by illumination and pose variations, several compensation techniques were proposed to overcome these issues. However, were successful for face recognition in partly lightened faces and not for facial expression recognition (FER). Attempts were made to implement FER. However these were not focused for intruder face recognition/monitoring. They lack the region of interest (ROI, in this case face detection) while processing, which is crucial for environment such as in a car (a possibility of another person behind/beside the driver). Thus, an Automated Video Surveillance system is presented in this paper. The system aims at tracking an object in motion and classifying it as a human or non-human entity, which would help in subsequent human activity analysis based on PCA based feature extraction.

Feature Extraction of Melanoma Data using Machine Learning Techniques

2021 ◽  
pp. 5352-5360
Author(s):  
R.Veeralakshmi, Dr.K.Merriliance
Keyword(s):  
Feature Extraction ◽  
Image Segmentation ◽  
Skin Diseases ◽  
Region Of Interest ◽  
Texture Features ◽  
Extraction Methods ◽  
Input Image ◽  
The Body ◽  
Machine Learning Techniques ◽  
Segmentation Technique

In our body the skin is the largest organ, it protects from injury, infection and also helps to maintain the temperature of the body. Melanoma Skin cancer is one of the most dangerous skin diseases and it is caused by an uncontrolled growth of abnormal skin cells, by ultraviolet radiation from sunshine. Melanoma is more common among white skins such as Americans than in darker skins. The digital lesion images have been analyzed based on image acquisition, pre-processing, and image segmentation technique. The image segmentation technique is applied to easily identify the affected portion in the skin input image. The images are enhanced using morphological filters and sharpen region of interest in an image, enhancement method enhanced the non-uniform background illumination and converts the input image into a binary image too easy to identify foreground objects. The mole of melanoma is segmented from the background using Active Contour algorithm. After that, the feature extraction methods such as Kernel PCA, SIFT are used to extract melanoma affected area in an image based on their intensity and texture features.

scholarly journals Breast Cancer Detection Using Combination of Feature Extraction Models

2018 ◽  
Vol 7 (3.12) ◽  
pp. 848
Author(s):  
T Suneetha Rani ◽  
S J Soujanya ◽  
Pole Anjaiah
Keyword(s):  
Breast Cancer ◽  
Feature Extraction ◽  
Cancer Detection ◽  
Texture Feature ◽  
Region Of Interest ◽  
Computation Time ◽  
Morphological Operations ◽  
Texture Feature Extraction ◽  
Mammogram Images

Recognition of either masses or tissues in a mammogram digital images is a key issue for radiologist. Present methods uses medial filter and morphological operations for detection of suspected cases in a mammogram. They use region of interest (ROI) segmentation for extraction of masses and classification of levels of severities.  Classification of large number of mammogram images based on breast cancer cases takes longer computation time for performing of ROI segmentation.  This is addressed by multi-ROI segmentation and it retrieves the textual properties of large mammogram images for effectively determining the breast cancer mammogram images.Experimental results shows the better performance of proposed method than existing ROI based texture feature extraction.

scholarly journals Efficient ConvNet Feature Extraction with Multiple RoI Pooling for Landmark-Based Visual Localization of Autonomous Vehicles

2017 ◽  
Vol 2017 ◽  
pp. 1-14 ◽  
Author(s):  
Yi Hou ◽  
Hong Zhang ◽  
Shilin Zhou ◽  
Huanxin Zou
Keyword(s):  
Feature Extraction ◽  
Autonomous Vehicles ◽  
High Efficiency ◽  
Region Of Interest ◽  
Fine Tuning ◽  
Visual Localization ◽  
Discrimination Power ◽  
Localization Accuracy ◽  
Feature Extraction Method ◽  
Convnet Feature

Efficient and robust visual localization is important for autonomous vehicles. By achieving impressive localization accuracy under conditions of significant changes, ConvNet landmark-based approach has attracted the attention of people in several research communities including autonomous vehicles. Such an approach relies heavily on the outstanding discrimination power of ConvNet features to match detected landmarks between images. However, a major challenge of this approach is how to extract discriminative ConvNet features efficiently. To address this challenging, inspired by the high efficiency of the region of interest (RoI) pooling layer, we propose a Multiple RoI (MRoI) pooling technique, an enhancement of RoI, and a simple yet efficient ConvNet feature extraction method. Our idea is to leverage MRoI pooling to exploit multilevel and multiresolution information from multiple convolutional layers and then fuse them to improve the discrimination capacity of the final ConvNet features. The main advantages of our method are (a) high computational efficiency for real-time applications; (b) GPU memory efficiency for mobile applications; and (c) use of pretrained model without fine-tuning or retraining for easy implementation. Experimental results on four datasets have demonstrated not only the above advantages but also the high discriminating power of the extracted ConvNet features with state-of-the-art localization accuracy.

Fabric defect detection with an attention mechanism based on hard sample training

2021 ◽  
pp. 004051752110600
Author(s):  
Hongge Yao ◽  
Qin Na ◽  
Shuangwu Zhu ◽  
Min Lin ◽  
Jun Yu
Keyword(s):  
Feature Extraction ◽  
Defect Detection ◽  
Data Augmentation ◽  
Region Of Interest ◽  
Attention Mechanism ◽  
Detection Accuracy ◽  
Convergence Region ◽  
Small Defect ◽  
Augmentation Strategy ◽  
Bounding Boxes

In view of the various types of fabric defects, and the problems of confusion, density unevenness and small target defects, which are difficult to detect, this paper builds a deep learning defect detection network incorporating an attention mechanism. The data augmentation strategy is used to enrich the number of samples of each defective type, and the enriched samples were extracted by the feature extraction network integrated with the attention mechanism, which can improve the feature extraction ability of confusable defect types and small defect types. Region proposal generation generates a proposal box for extracted features, and adds an online hard example mining strategy to re-learn hard examples to accelerate network convergence. Region feature aggregation maps the proposal box to the feature map to obtain the region of interest. Finally, the defect features are classified and the bounding boxes are regressed. The results show that this algorithm can effectively detect 39 categories of fabric defects with a detection speed of 0.085 s and a detection accuracy of 0.9338.

Situation-Oriented Hierarchical Classification for Sightseeing Images Based on Local Color Feature

2013 ◽  
Vol 17 (3) ◽  
pp. 459-468 ◽  
Author(s):  
Chia-Huang Chen ◽  
◽  
Yasufumi Takama
Keyword(s):  
Feature Extraction ◽  
Hierarchical Classification ◽  
Region Of Interest ◽  
Hierarchical Organization ◽  
Night Time ◽  
Local Color ◽  
Color Feature ◽  
Second Stage ◽  
Color Features ◽  
Three Stages

Nowadays, tourists take lots of photos and share them on album websites, so the meaningful grouping of images becomes important and useful. Specifically, sightseeing scenes vary with different situations such as weather and season. The categorization of different situations is thus expected to be beneficial to tourists planning when to visit different places. This paper proposes a hierarchical classification method based on local color feature extraction from the designed region of interest (ROI) andK-means clustering to categorize sightseeing images into several meaningful situations. Hierarchical organization consists of three stages and four situations. In the first stage, night-time images are discriminated from daytime images, then daytime images are divided into sunrise/sunset and other images in the second stage. Finally, cloudy images are separated from sunshiny images in other images obtained in the second stage. Experimental results show that the extraction of color features within the ROI is effective in obtaining clusters with high precision and recall.

Sign in / Sign up

Export Citation Format

Share Document