Leaves, insect wings, and fish fins are only a few examples of flexible structures found in nature that come in a myriad of different shapes and sizes, which might affect the way they interact with the surrounding flow. This observation inspired us to study how the planform shape of cantilevered flexible sheets or flags affects their flapping dynamics, critical flapping velocity, and the aerodynamic forces they experience. Here, we selected a shape design space inspired by leaves. A database is created using original leaf-shapes from a public repository that were digitised, and modified into symmetrical shapes. The main geometrical features of the leaf database are extracted using a principal component analysis of the discretised leaf contours. New leaf-inspired shapes are generated as linear combinations of the first seven modes using sampled coefficients. The flapping deformation of these new shapes are measured during a series of automated experiments. To characterise the flapping dynamics of the flags, we gathered spatially and temporally resolved deformation data of the flapping leaf at mid-height using an event-based camera. To further reduce the dimensionality of the data and characterise the flapping dynamics, we spatially cluster the data in the POD feature space and calculate the transition probabilities between clusters following a cluster-based network modelling approach. In the video Dancing with the wind, learning from the choreography of flapping leaves, we make an analogy between the flapping dynamics of flags and a dance routine, as both can be presented by the succession of discrete steps that are repeated in time.