no code implementations • 1 Jan 2021 • Arash Akbarinia, Raquel Gil-Rodriguez, Alban Flachot, Matteo Toscani
Our results show, with respect to the baseline network (whose input and output are RGB) 5-10% higher classification accuracy is obtained with decorrelating ColourConvNets.
1 code implementation • 28 Dec 2020 • Alban Flachot, Arash Akbarinia, Heiko H. Schütt, Roland W. Fleming, Felix A. Wichmann, Karl R. Gegenfurtner
High levels of color constancy were achieved with different DNN architectures.
1 code implementation • 30 Sep 2020 • Arash Akbarinia, Raquel Gil-Rodríguez, Alban Flachot, Matteo Toscani
We also observed that the learnt embedding space is easier to interpret in colour opponent models.