TY  - CONF
AU  - Marco Buzzelli
AU  - Joost Van de Weijer
AU  - Raimondo Schettini
A2  - ICIP
PY  - 2018//
TI  - Learning Illuminant Estimation from Object Recognition
BT  - 25th International Conference on Image Processing
SP  - 3234
EP  - 3238
KW  - Illuminant estimation
KW  - computational color constancy
KW  - semi-supervised learning
KW  - deep learning
KW  - convolutional neural networks
N2  - In this paper we present a deep learning method to estimate the illuminant of an image. Our model is not trained with illuminant annotations, but with the objective of improving performance on an auxiliary task such as object recognition. To the best of our knowledge, this is the first example of a deeplearning architecture for illuminant estimation that is trained without ground truth illuminants. We evaluate our solution on standard datasets for color constancy, and compare it with state of the art methods. Our proposal is shown to outperform most deep learning methods in a cross-dataset evaluationsetup, and to present competitive results in a comparison with parametric solutions.
L1  - http://refbase.cvc.uab.es/files/BWS2018.pdf
UR  - http://dx.doi.org/10.1109/ICIP.2018.8451229
N1  - LAMP; 600.109; 600.120
ID  - Marco Buzzelli2018
ER  -