PCS21, Diss, Transalp, RDPlot, and PyCabac

Since my last post in June 2021 a lot has happened, and I will try to give a general update on some of my projects. So let's jump right into it…

We received the best student paper award for the work with respect to "Sparse Coding-based Intra Prediction in VVC" at PCS21. I think, it was a great appreciation for all the effort, which we put into this intra prediction method. I definitely want to thank once again my colleagues and co authors Dominik Mehlem, Maria Meyer, and Christian Rohlfing for all their support. IMHO, only great teams can make great achievements, and this is a great team.

In early July 2021 I handed in my dissertation with the working title "Sparse Signal Modelling in Video Coding". As you can guess from the title and my former publications, the thesis deals with entry points for sparse signal models into the hybrid video coding scheme.

In July 2021, I crossed the alps with my road bike, which was extraordinary great. One day, I was frozen to the bone in the snow at passo dello Stelvio, and the other day I had summer feelings at lago di Garda. It was a ride under unstable weather conditions and in parts a little tough. However, we made it, and it was a great experience.

RDPlot is a tool for plotting Rate Distortion (RD) curves and calculating Bjontegaard Delta (BD) rate changes. The project was started roughly 4 years ago, and made some progress over the past years. Recently, the tool raised attention at JVET-meetings, and some feature requests were forwarded to me. With Mirco Dilly's help, we were able to improve the tool and add some of the requested features. Most notably, plotting RD-curves with Confidence Intervals (Cis) is possible now, and the calculation of BD-rate changes is adjustable with respect to the CIs. I think, that is great news, as the feature set of RDPlot has proven to be still growing.

Context-based Adaptive Binary Arithmetic Coding (CABAC) is the state-of-the-art entropy coding method in modern video coding standards, such as High Efficiency Video Coding (HEVC) and Versatile Video Coding (VVC). As video coding standards evolve over the generations, also the probability modelling in CABAC does. For the CABAC engine in HEVC, a MATLAB port already existed, which allows for rapid prototyping of entropy coding methods for arbitrary binary sources. For the CABAC engine in VVC, no such port was published so far. As rapid prototyping is also performed in Python in many cases, we decided to implement a Python port this time, namely PyCabac. If you are interested, go ahead and check out the repository. From a user perspective, it should be plug and (entropy) code :-).