Fig. 2

Architecture of the deep convolutional mixture density network. This architecture directly uses image matrices as input. In this figure, three convolutional layers are drawn, showing the local connection between the different layers. In the fully connected part, a MDN with one hidden layer is used and a vector for each parameter of the GMM is given as result (μ, σ, ω). A sample of a predicted PDF is shown exemplarily.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.