| CPC H04N 19/70 (2014.11) [H04N 19/136 (2014.11); H04N 19/42 (2014.11); H04N 19/91 (2014.11)] | 14 Claims |

|
1. A method to decode with a processor a coded image or a coded video sequence with a neural network, the method comprising:
receiving a coded image or a coded video sequence comprising coded pictures and syntax parameters for an entropy model for latent features of the coded image or coded video sequence, wherein the entropy model comprises one or more probability density functions (PDFs) selected from a list of available PDFs that includes different distribution types, wherein the list of available PDFs comprise a Gaussian PDF, a Laplace PDF, and a uniform PDF;
parsing the syntax parameters for the entropy model for the latent features to generate model parameters for the entropy model, wherein the syntax parameters comprise: the number of one or more PDFs being used, an identifier of each PDF being used among the list of available PDFs; and
decoding, using the neural network, the coded image or the coded video sequence using the generated model parameters for the entropy model.
|