Inspection of rice seeds is a crucial task for plant nurseries and farmers since it ensures seed quality when growing seedlings. Conventionally, this process performed by expert inspectors who manually screen large samples to identify their species assess the cleanness batch. In quest automate screening through machine vision, variety approaches utilise appearance-based features extracted from RGB images while others spectral information acquired using Hyperspectral Imaging (HSI) systems....

Recently, the recent advancement of deep learning with capacity to perform automatic high-level feature extraction has achieved promising performance for sensor-based human activity recognition (HAR). Among different methods, Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) have been widely adopted. However, scalar outputs pooling in CNN only allow get invariance but not equivariance. The capsule networks (CapsNet) vector output routing by agreement is able capture In...

This paper presents a system for automated classification of rice variety seed production using computer vision and image processing techniques. Rice seeds different varieties are visually very similar in color, shape texture that make the at high accuracy challenging. We investigated various feature extraction techniques efficient representation. analyzed performance powerful classifiers on extracted features finding robust one. Images six northern Vietnam were acquired analyzed. Our...

Abstract Skeleton‐based human action recognition has emerged recently thanks to its compactness and robustness appearance variations. Although impressive results have been obtained in recent years, the performance of skeleton‐based methods be improved deployed real‐time applications. Recently, a lightweight network structure named Double‐feature Double‐motion Network (DD‐Net) proposed for recognition. With high speed, DD‐Net achieves state‐of‐the‐art on hand body actions. The could not...

Automatic classification and disease detection in medical images, aided by machine learning, provide crucial support to prevent overlooked instances ensure prompt treatment of diseases. Despite impressive achievements the field polyp from endoscopic other diseases, such as reflux esophagitis, esophageal cancer, gastritis, gastric duodenal ulcer, is still subject significant limitations remains a challenging area study because their different more characteristics. This paper proposes method...

Abstract Emotion recognition is a critical component of human-computer interaction, facilitating more natural and effective communication between users machines. This study explores the positive negative emotions using multichannel Electroencephalogram (EEG) analysis approach, incorporating spectral entropy machine learning techniques. Spectral entropy, which quantifies complexity irregularity EEG signals, employed to extract relevant features from raw data. These are then fed into various...

Biological methods have been recently found to be eco-friendly and cost-effective alternatives for synthesizing silver nanoparticles (AgNPs). This study highlights a green process wherein AgNPs were synthesized coated by seven leaf extracts consisting of Piper betle L., sarmentosum Roxb, Averrhoa carambola Syzygium aqueum (Burn.f.) Alstion, nervosum A. Cunn. ex DC, Psidium guajava Couroupita surinamensis Mart. Ex Berg under the acceleration ultrasound. The optimized characterized UV–visible...

In this paper we investigate a special type of denial service (DoS) attack on 802.11-based networks, namely deauthentication/disassociation attack. the current IEEE 802.11 standards, whenever wireless station wants to leave network, it sends deauthentication or disassociation frame access point. These two frames, however, are sent unencrypted and not authenticated by Therefore, an attacker can launch DoS spoofing these messages thus disabling communication between device its We propose...

A bi-modal emotion recognition approach is proposed for of four emotions that integrate information from gestures and speech. The outputs two unimodal systems based on affective speech expressive gesture are fused a decision level fusion by using weight criterion best probability plus majority vote methods, the performance classifier which performs better than each uni-modal helpful in recognizing suitable communication situations. To validate proposal, fifty Japanese words (or phrases) 8...

In this work, we study the ability to use hand gestures for human-machine interaction from wrist-worn sensors. Towards goal, design a prototype capture RGB video stream of gestures. Then built new gesture dataset (named WiGes) with various subjects in home appliances different environments. To best our knowledge, is first benchmark released studying camera. We then evaluate CNN models vision-based recognition. Furthermore, deeply analyze that produce trade-off between accuracy, memory...

This paper describes a fusion technique for species identification from images of different plant organs. Given series image organs such as branch, entire, flower or leaf, we firstly extract confidence scores each single organ using state-of-the-art deep convolutional neural network (CNN). After that, deploy various schemes the approaches including not only conventional transformation-based (sum rule, max product rule) but also classification-based approach (support vector machine). Then...

Rice is one of the most cultivated cereal in Asian countries and Vietnam particular. Good seed germination important for rice quality, that impacts production crop yield. Currently, evaluation carried out manually by experienced persons. This a tedious time-consuming task. In this paper, we present system automatic rate based on advanced techniques computer vision machine learning. We propose to use U-Net - convolutional neural network segmentation separation seeds. Further processing such...

A conventional method to inspect the varietal purity of rice seeds is based on human visual inspection where a random sample drawn from batch. This tedious, laborious, time consuming and extremely inefficient task. paper presents an automatic seed using Hyperspectral imaging machine learning, automatically detect unwanted other varieties which may be contained in image data Near-infrared (NER) camera are acquired for six common varieties. The results applying two classifiers presented,...

This paper presents a plant identification method from the images of simple leaf with complex background. In order to extract image, we firstly develop an interactive image segmentation for mobile device tactile screen. allows separate region background in few manipulations. Then, kernel descriptor build representation. Since may be taken at different scale and rotation levels, propose two improvements extraction that makes robust rotation. Experiments carried out on subset ImageClef 2013...

In this paper a navigational aid for visually impaired people is presented. The system uses RGB-D camera to perceive the environment and implements self-localization, obstacle detection classification. novelty of work threefold. First, self-localization performed by means novel tracking approach that both depth color information. Second, provide user with semantic information, obstacles are classified as walls, doors, steps residual class covers isolated objects bumpy parts on floor. Third,...