One of the main goals of this study is to create a benchmark dataset for detection tasks for forest cover changes. Even though there is a smaller number of datasets available for change detection almost all of them are focusing on urban change detection. Here in this study, we have collected a dataset consisting of three band colour (RGB) images for change detections of forest covers all over the world. All collected images are in identical width and height which are 970px and 1980px respectively. Our main source for collecting images was the Google Earth historical image feature. It provides historical images of certain geographical locations back to the 1980s. However, some areas did not have images in every time step and the interval between each capture also varied. Some areas for example amazon forest have most of the images in those time steps. But when considering South Asia, particularly Sri Lanka there are not much of historically captured images. So one question we faced when collecting data was how to choose locations where we could get most images. Also, what are the criteria we gonna use when selecting time intervals between captured images? So the first question was sold by using the 2004 to 2017, Deforesting Fronts report. There are 12 fronts where we can identify fronts where deforesting is happening. The approach we used to solve the second question is by defining a year (365 days) as the time interval between two captures over a certain location.