Zobrazeno 1 - 5
of 5
pro vyhledávání: '"Hedjazi, Mohamed Abbas"'
Although recent deep learning-based calibration methods can predict extrinsic and intrinsic camera parameters from a single image, their generalization remains limited by the number and distribution of training data samples. The huge computational an
Externí odkaz:
http://arxiv.org/abs/2212.02379
Autor:
Hedjazi, Mohamed Abbas, Genc, Yakup
Publikováno v:
Knowledge-Based Systems, Volume 217, 6 April 2021, 106789
Recent GAN-based (Generative adversarial networks) inpainting methods show remarkable improvements and generate plausible images using multi-stage networks or Contextual Attention Modules (CAM). However, these techniques increase the model complexity
Externí odkaz:
http://arxiv.org/abs/2009.14721
Autor:
Hedjazi, Mohamed Abbas, Genc, Yakup
Image inpainting is one of the most challenging tasks in computer vision. Recently, generative-based image inpainting methods have been shown to produce visually plausible images. However, they still have difficulties to generate the correct structur
Externí odkaz:
http://arxiv.org/abs/2002.09280
Autor:
Hedjazi, Mohamed Abbas, Genc, Yakup
Publikováno v:
In Knowledge-Based Systems 6 April 2021 217
Autor:
Hedjazi, Mohamed Abbas, Genc, Yakup
Publikováno v:
In Signal Processing: Image Communication April 2021 93