2 results
Search Results
Now showing 1 - 2 of 2
Conference Object Citation - Scopus: 22Kinshipgan: Synthesizing of Kinship Faces From Family Photos by Regularizing a Deep Face Network(IEEE Computer Society, 2018) Ozkan,S.; Ozkan,A.In this paper, we propose a kinship generator network that can synthesize a possible child face by analyzing his/her parent's photo. For this purpose, we focus on to handle the scarcity of kinship datasets throughout the paper by proposing novel solutions in particular. To extract robust features, we integrate a pre-trained face model to the kinship face generator. Moreover, the generator network is regularized with an additional face dataset and adversarial loss to decrease the overfitting of the limited samples. Lastly, we adapt cycle-domain transformation to attain a more stable results. Experiments are conducted on Families in the Wild (FIW) dataset. The experimental results show that the contributions presented in the paper provide important performance improvements compared to the baseline architecture and our proposed method yields promising perceptual results. © 2018 IEEE.Conference Object Citation - Scopus: 24-Stage Target Detection Approach in Hyperspectral Images(IEEE Computer Society, 2018) Ozdil,O.; Gunes,A.; Esin,Y.E.; Ozturk,S.; Demirel,B.Practical target detection systems require an automatic way to detect targets with high accuracy. Detection errors is not tolerable and they should be reduced as much as possible. In classical detection systems, generally single target detection algorithm is performed and the result will be evaluated according to the thresholding techniques. However, in these uncontrolled systems, false alarm rate strongly depends on the thresholding technique success. It is very hard to find a general and constant threshold value for images taken at different conditions and practical detection systems needs reliable threshold value. In this paper, we propose a new multi-stage target detection system which is the combination of different detection algorithms and thresholding technique. This system compose of 4-stages, i.e. namely 1-initial target detection (ACE, GLRT), 2-adaptive Constant False Alarm Rate (CFAR) thresholding, 3-spatially grouping, 4-statistical confidence operation. This system configuration removes the need for interactive user and it automatically implements confirmation and rejection steps. Moreover, this system can be used both for pure pixel and subpixel target detection purposes and it reduces computational processing time considerably with the implementation of consequtive processing stages. © 2018 IEEE.
