In this technical report, we present our solution to the Traffic4Cast 2021 Core Challenge, in which participants were asked to develop algorithms for predicting a traffic state 60 minutes ahead, based on the information from the previous hour, in 4 different cities. In contrast to the previously held competitions, this year's challenge focuses on the temporal domain shift in traffic due to the COVID-19 pandemic. Following the past success of U-Net, we utilize it for predicting future traffic maps. Additionally, we explore the usage of pre-trained encoders such as DenseNet and EfficientNet and employ multiple domain adaptation techniques to fight the domain shift. Our solution has ranked third in the final competition.

However, because Cellular automata (CA) is a widely used modeling theory. of the the constraint of computing power, and the limited From the perspective of physics, CA refers to a dynamic system defined in a cell space composed of cells with discrete and finite states, which evolved in discrete time dimensions according to certain local rules. Cells are the most basic component of CA which are distributed in discrete Euclidean space positions. Each cell in the lattice grid takes from a finite set of discrete states, follows the same local rules of actions, and updates simultaneously according to the rules. Other cells within the local space which may interact with the rules are defined as the "neighborhood". While the evolution for each cell only take place based on local information, a large number of cells make the evolution of the entire dynamic system happen through interactions, and hence form a dynamic effect globally. CAs are not determined by strictly defined equations or functions, but are constituted by Figure 1: Von Neumann neighborhood (red) and Moore a series of rules for constructing models. Therefore, CA is a neighborhood (blue).