MSP-CNN Screen Content Video and Train Dataset

In order to accelerate the CU partitioning in VVC-SCC, we have established a specialized dataset for screen content. Firstly, we collect a wide variety of screen content videos and images in YUV format. Secondly, we encode these sequences under four different Quantization Parameters (QPs) to obtain comprehensive information about CU partitioning in VVC-SCC. This includes the length, width, position, mode labels, and RDcost values corresponding to each mode. This dataset will provide valuable data support for optimizing the CU partitioning process in screen content coding.