| Titre : | Technique using deep learning for the selection of perceptually salient objects |
| Auteurs : | CHAHLA BELBAHI, Auteur ; Mohamed Chaouki Babahenini, Directeur de thèse |
| Type de document : | Monographie imprimée |
| Editeur : | Biskra [Algérie] : Faculté des Sciences Exactes et des Sciences de la Nature et de la Vie, Université Mohamed Khider, 2021 |
| Format : | 1 vol. (59 p.) / ill. / 29 cm |
| Langues: | Anglais |
| Résumé : | Visual perception is the impression captured by the eyes, which models the most relevant information in any image. Its purpose is to choose which area of the image will be analyzed before the others because it would be potentially more interesting. Thus it reduces the amount of information to be processed, and therefore accelerates the entire process of vision.In this dissertation, we proposed to study and implement a "visual detection with deep learning" perception model to produce a salient greyscale 2D image that will be used in posterior processing. image enhancement, which will allow to make treatments according to the importance of the visual perception representation leading to find the right compromise between the good quality of the treatments and the calculation time of the result image. |
| Sommaire : |
General Introduction 7
Chapter 01 : Visual perception and color 1. Introduction 10 2. Neurology Context of Anticipated Vision and Attention 10 3. Human Visual System (HVS) 10 3.1.The Structure of the Eyes 11 3.2. Components of the eye 11 3.2.1. Sclera 11 3.2.2. Choroid 11 3.2.3. The retina 11 4. Visual perception 12 5. Color perception 12 5.1.Color 12 6. Color Systems 13 6.1. RGB System 14 6.2. CIE XYZ System 16 6.3.TSL /TSV System (Teinte Saturation Luminance /Teinte Saturation 17 6.4. The absolute CIE L * a * b 1976 color space . 17 7. Conclusion 18 Chapter 02 : History of salient object detection (Saliency Models) 1. Introduction 20 2. Visual attention 21 2.1.Psychophysics 21 2.2.Neurophysiology 21 2.3. Computational Modeling 21 2.3.1. Classic Models 22 2.3.1.1Top Down Models 22 2.3.1.2 Buttom-up Models 2.3.2 Neural models 23 3. The saliency Map 24 4. Conclusion 25 Chapter 03 :Deep learning for visual saliency detection. 1. Introduction 27 2. Visual saliency detection using deep CNN .27 2.1 Implementation of fully convolution network 29 2.2 Refinement of attention map 30 2.3 Gaussian filter 31 3. conclusion 32 Chapter 04 : Implementation and results. I. Introduction 34 2. Development tool 34 2.1. Programming languages 34 2.2. Libraries used .35 2.2.1. OpenCv .35 2.2.2. Keras 35 3. Implementation 36 3.1. Information about Dataset 36 3.2. Organizing Data into Folder .36 3.3. Pre-processing Function and Generator 37 3.4. Defining Model and Loss Function .39 3.5. Defining Custom Loss Function derived from MSELoss 40 3.6. Visulazing Model output and saving Model Weights 40 3.7. Saving Model 41 4. Evausation .41 5. Results 42 6.Conclusion 46 General Conclusion .47 Bibliography and webography ..48 |
| Type de document : | Mémoire master |
Disponibilité (1)
| Cote | Support | Localisation | Statut |
|---|---|---|---|
| MINF/675 | Mémoire master | bibliothèque sciences exactes | Consultable |
