Santanu Koley

A aplica o Android de localiza o de ve culos apresentada neste estudo tira partido do poder da tecnologia m vel moderna e dos servi os baseados na localiza o para fornecer uma solu o intuitiva e eficiente para a localiza o de ve culos em tempo real. Com a utiliza o generalizada de smartphones e a ubiquidade da tecnologia GPS, esta aplica o oferece um conjunto de ferramentas vers til para monitorizar e gerir os movimentos dos ve culos. O principal objetivo desta aplica o dar aos utilizadores a capacidade de seguir a localiza o em tempo real dos ve culos utilizando os seus dispositivos Android. Ao aproveitar as capacidades do GPS, a aplica o oferece actualiza es em tempo real das posi es dos ve culos, permitindo aos utilizadores monitorizar os seus bens, a sua frota ou o paradeiro dos seus entes queridos. Al m disso, a aplica o armazena dados hist ricos de localiza o, facilitando a an lise e reprodu o de rotas para uma tomada de decis o informada. Este estudo n o s mostra a implementa o t cnica da localiza o em tempo real, como tamb m sublinha as potenciais aplica es em v rios dom nios, incluindo o transporte pessoal, a log stica, os servi os de emerg ncia, entre outros.

HAR est un domaine de recherche li la d tection volontaire d'activit s quotidiennes effectu es par des personnes sur la base de s ries de donn es temporelles utilisant des capteurs. Le HAR couvre diff rents domaines tels que la surveillance, le suivi des b b s, les soins de sant aux personnes g es, les voitures intelligentes, en utilisant diff rentes approches pour r soudre les probl mes de mani re efficace et pr cise. Les syst mes HAR traditionnels utilisent des capteurs portables tels que des unit s de mesure inertielle (IMU) et des capteurs d' tirement pour reconna tre l'activit .Cette approche donne des r sultats remarquables pour les activit s de base de l'utilisateur, telles que s'asseoir, se tenir debout et marcher. Mais pour les activit s complexes telles que la course, le saut, la lutte et le balancement, les syst mes HAR bas s sur des capteurs pr sentent des taux de classification erron e plus lev s en raison des erreurs de lecture des capteurs. Ces erreurs de capteur entra nent les pires r sultats de classification possibles et r duisent les performances globales du syst me HAR. En utilisant une combinaison de CNN et de LSTM, les donn es seront extraites et trait es partir de vid os. Dans cet ouvrage, un r seau neuronal convolutionnel profond est propos , qui permet d'extraire les caract ristiques pour la collecte des donn es partir de la s quence d'entr e (vid o). Ensuite, la LSTM sera utilis e pour d terminer les relations temporelles entre les images.

Edge detection is very important terminology in image processing and for computer vision. Edge detection is in the forefront of image processing for object detection, so it is crucial to have a good understanding of edge detection operators. An improved scheme for contour detection with better performance measure has been proposed in this book with the help of image gradients. A 9x9 Laplacian and Gaussian (LOG) filter has been proposed. The present study has shown that the gradient images obtained by the 9x9 Laplacian and Gaussian (LOG) mask appears to be much clearer with sharp and prominent edges than those obtained through 5x5 Laplacian and Gaussian (LOG) filter and also provides better results than traditional edge detectors. The method has been applied to a number of real life digital images and better performance measure of contour detection has been achieved. Comparative analyses of different edge detection operators in image processing are also presented and discussed. It has been observed that the performance of canny edge detection operator is much better then Sobel, Roberts, Prewitt and Laplacian of Gaussian (LoG) in respect to the image appearance and object boundary.