Abstract: A computer vision processing system is provided. The system includes one or more target devices and a processing unit. The target devices are configured to run the executable code of an image processing pipeline. The processing unit is configured to receive a series of application programming interface (API) calls and create a raw graph accordingly, redraw the raw graph into a compilable graph by sequentially processing each node, and compile the compilable graph into the executable code of the image processing pipeline. The series of API calls includes at least one tiling API call to set at least one of the nodes and at least one of the data objects as tileable. Each tileable node corresponds to multiple parallel processing nodes in multiple branches in the compilable graph, and each tileable data object corresponds to multiple tile data objects in the branches in the compilable graph.