Gutter and Bollard Datasets

The gutter dataset contains 85 images of the uncovered gutter object and their annotations, while the bollard dataset has 150 images of multiple bollard objects and their annotations. The gutter is 245 x 326 pixels in size, while the bollards are 400 x 300 pixels, each bollard image has multiple items. The resolution of each image contained in the gutter and bollard dataset is 96dpi. The text files hold the coordinate values of each object (<class id> <Xo/X> <Yo/Y> <W/X> <H/Y>), which represents the predefined bounding box during the annotation. The dataset would be suitable for image recognition researches.