Contrast evaluation of butterflies in ecological images by Quantum mechanics
Description
We present the results obtained by our algorithm which adapt wave functions in the Schrodinger equation to be deployed in a multi-layered graph, on the the public Leeds Butterfly dataset (J. Wang, K. Markert, M. Everingham, Learning models for object recognition from natural language descriptions. Proc. br. Mach. Vis. Conf., (2009), pp. 2.1-2.11, https://doi.org/10.5244/C.23.2.). It contains 832 ecological photographs of ten species of butterflies, with a distribution of 55 to 100 images for each specie. These images are submitted to several artifacts as the shadow effect, the high diversity of background and butterfly patterns, presence of distractors and the high similarity of some butterfly and background regions. Our results correspond to contrast maps that highlight butterfly regions and discard background areas. The ground truth images are provided to evaluate visually the performance of our algorithm. Folders descriptions: -> Images: Color ecological images. -> Ground Truth: The real position of butterflies in the images. -> Contrast Maps: Results returned by our algorithm. -> Contrast Estimation: a compressed folder containing the above folders