Informatique

Nombre total de résultats : 3
Pertinence Titre A-Z Plus récents Plus anciens
10 25 50
Année de publication
et

A Survey on Lightweight CNN-Based Object Detection Algorithms for Platforms with Limited Computational Resources

BOUGUETTAYA Abdelmalek, Kechida Ahmed, TABERKIT Mohammed Amine  (2019)
Publication

Autonomous drones must be able to identify the existence of one or more objects of interest in a complex environment with high accuracy and speed to fly around safely. Most existing object detection techniques, based on traditional machine learning algorithms, can't offer acceptable performance in complicated environments. Deep Convolutional Neural Networks (CNNs) provide us such ability with high performance. Today, deep CNN-based object detection algorithms are more and more used in Artificial Intelligence (AI) applications. However, it still very difficult to deploy large CNNs architectures on small devices with limited hardware resources, because they consist of millions of parameters, which make them computationally very exhausting. Lightweight CNN architectures are proposed as a solution to make the deployment of deep neural networks on small devices feasible. This paper focuses on reviewing recent used lightweight CNN architectures that can be implemented on embedded targets to improve the object detection performance for small devices-based systems, like drones. We need to select fast and lightweight CNN models to use them on drone platforms. The purpose of this reviewing is to choose the most accurate and fastest algorithm to implement it on our drones. Voir les détails

Mots clés : Computer vision, Deep Learning, Object Detection, Convolutional Neural Network, lightweight CNN

Lightweight CNNs-Based Object Detection forEmbedded Systems implementation

BOUGUETTAYA Abdelmalek, Kechida Ahmed, TABERKIT Mohammed Amine  (2019)
Article de conférence

Deep Learning algorithms, based on the implementation of Convolutional Neural Networks (CNN), are more and more used in Artificial Intelligence (AI) applications, especially in the image recognition field, like image classification, object detection, segmentation. These algorithms learn from training data a set of parameters to create a model, which is capable of performing a classification task with high accuracy. The most recent models consist of millions of parameters, which make it computationally very exhausting, especially in the field of embedded systems where resources are very limited. Recently, deep learning and computer vision are highly used to realize a fully-autonomous drone and self-driving cars, which does not need human intervention. Computer vision is a field focused on enabling drones to interpret and understand the content of an image or a video using CNNs. This paper focuses on reviewing recent lightweight CNNs architectures used that can be implemented on embedded targets. Voir les détails

Mots clés : Computer vision, Deep Learning, Object Detection, Convolutional Neural Network, lightweight CNN

Software Features Extraction From Object-Oriented Source Code Using an Overlapping Clustering Approach

Imad Eddine ARAAR, Hassina SERIDI  (2016)
Publication

For many decades, numerous organizations have launched software reuse initiatives to improve their productivity. Software product lines (SPL) addressed this problem by organizing software development around a set of features that are shared by a set of products. In order to exploit existing software products for building a new SPL, features composing each of the used products must be specified in the first place. In this paper we analyze the effectiveness of overlapping clustering based technique to mine functional features from object-oriented (OO) source code of existing systems. The evaluation of the proposed approach using two different Java open-source applications, i.e. “Mobile media” and “Drawing Shapes”, has revealed encouraging results. Voir les détails

Mots clés : feature model, software product line, overlapping clustering, reverse engineering, program analysis