Understanding crowd behavior is a fundamental scientific challenge that can be used, e.g., to analyze diffusion of ideas or viruses, design safety measures for urban areas, and to understand dynamics of collective behavior. Pixel-based camera counting methods provide an interesting, privacy preserving and low-cost solution for crowd monitoring as, e.g., CCTV cameras are already commonplace. However, there are several challenges when monitoring changes in these images: the background can change due to weather or lighting condition changes, the angle of the camera and the distance of objects affect the size of objects in the visual feed, and so forth. The objective of the Thesis is to conduct a literature survey on different pixel-based crowd counting methods and to compare their performance in determining the level (or intensity) of crowd activity.