You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi,
CompressAI use ranscoder for entropy coding with zero-mean gaussuan, e.g. for encoding $y$ with $N( \mu, \sigma^2)$, ranscoder encodes $Q(y - \mu)$ with $N(0, \sigma^2)$.
I wonder if ranscoder can encode $Q(y)$ with $N( \mu, \sigma^2$) so that encoder/decoder mismatch can be resolved especially in some autoregressive coding schemes.
Of course, in the case of JointAutoregressiveHierarchicalPriors, it is possible to train well with the current coding scheme, but in my case, as I apply the autoregressive coding scheme to the video compression model, error propagation occurs severely.
As my knowledge, tfc (tensorflow compression) support this operation.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Hi,$y$ with $N( \mu, \sigma^2)$ , ranscoder encodes $Q(y - \mu)$ with $N(0, \sigma^2)$ .
CompressAI use ranscoder for entropy coding with zero-mean gaussuan,
e.g. for encoding
I wonder if ranscoder can encode$Q(y)$ with $N( \mu, \sigma^2$ ) so that encoder/decoder mismatch can be resolved especially in some autoregressive coding schemes.
Of course, in the case of
JointAutoregressiveHierarchicalPriors
, it is possible to train well with the current coding scheme, but in my case, as I apply the autoregressive coding scheme to the video compression model, error propagation occurs severely.As my knowledge, tfc (tensorflow compression) support this operation.
Can i implement this operation with CompressAI?
Thank you in advance.
Beta Was this translation helpful? Give feedback.
All reactions