User Tools

Site Tools



This shows you the differences between two versions of the page.

Link to this comparison view

Next revision
Previous revision
projects:year4:15.9 [2019/08/21 12:40]
sally.johnson created
projects:year4:15.9 [2021/06/02 14:40] (current)
sally.johnson [Table]
Line 1: Line 1:
-====== 15.9 - Learn to Segment ======+====== 15.9 - Learn to Segment by Operational Object Segmentation Networks======
 ===== Project - Summary ===== ===== Project - Summary =====
 <WRAP leftalign box > <WRAP leftalign box >
-Automatic image content description is vital problems in computer vision that artificial intelligence and natural language processingThe primary challenge towards this goal is in the design of a multi-model approach that is rich enough to aim simultaneously about contents of images and their representation in terms of words or sentencesWe present a multi-model approach based on a deep learning architecture that combines recent advances in computer vision such as; salient object proposal prediction, and object detection to generate natural sentences describing an image. Leveraging recent advances in recognition of objects, their attributes and locations, however they are limited in their expressivity. Moreover, current object detection methods still suffer various problems in localization and processing time that render them unreliable and inadequate as they are still slow at test time. We target the high-level goal of annotating the contents of images based salient regions or segments of images and study the multimodal correspondence between words and images. The idea is to correctly labeling scenes, objects and regions with a fixed set of categories, while our focus is on richer and higher-level descriptions of regions. The proposed approaches can also be used in text to image search in large scale image retrieval systems.+**Objectives:** 
 +  * To develop novel methods for supervised salient object segmentation task that provides many industrial applications such as camera auto-focusing, advertisement assessment, etc. 
 +  * To provide general trainable segmentation systems to the industrial partners such that each partner can train its own network in accordance with their application of interest. 
 +  * To research novel Neural Network structures and to investigate the feasibility and effectiveness of the proposed models for salient object segmentation task.
 </WRAP> </WRAP>
 ===== Project - Team ===== ===== Project - Team =====
-^ Team Member           ^ Role         ^ Email          ^ Phone Number   ^ Academic Site/IAB   ^ +^ Team Member      ^ Role         ^ Email                    ^ Phone Number     ^ Academic Site/IAB   ^ 
-| Moncef Gabbouj        | PI           Not available  Not available  | Tampere University +| Moncef Gabbouj   | PI    +358 400 736613  | Tampere University 
-| Serkan Kiranyaz       | PI           | Not available  | Not available  | Tampere University +| Serkan Kiranyaz  | PI   97 43 063 5600   | Tampere University 
-| Iftikhar Ahmad        | PI           Not available  Not available  | Tampere University +Caglar Aytekin   | PhD Student  | Not available            | Not available    | Tampere University  |
-Alexandros Iosifidis  | PI           | Not available  | Not available  | Tampere University +
-| Muhammad Adeel Waris  | PhD Student  | Not available  | Not available  | Tampere University  |+
Line 16: Line 18:
 ===== Project - Impact and Uses/Benefits ===== ===== Project - Impact and Uses/Benefits =====
 <WRAP leftalign box > <WRAP leftalign box >
-Industrial partner can train this system for any object classification task that it desires. Such systems can serve the companies that are dealing with computer vision problems such as camera smart object auto-focusing, advertisement assessment, face detection, and many other applications basically including any object/region detection task.+The Convolutional Segmentation Networks and Convolutional Kernel Networks combined with Quantum Cuts provide two trainable segmentation networks that have the capacity to achieve state-of-the-art performance in salient object segmentation. The impact of such a generalized segmentation network is the fact that, given the ground truths, an industrial partner can train this system for any segmentation task that it desires. Such systems can serve the companies that are dealing with computer vision problems such as camera auto-focusing, advertisement assessment, face segmentation and many other applications basically including any object/region segmentation task.
 </WRAP> </WRAP>
Line 24: Line 26:
 ===== Project - Documents ===== ===== Project - Documents =====
projects/year4/15.9.1566409215.txt.gz · Last modified: 2019/08/21 12:40 by sally.johnson