Due to distortion, noise, segmentation errors, overlap, and occlusion of objects in digital images, it is usually impossible to extract complete object contours or to segment the ...
The main difficulty in the binary object classification field lays in dealing with a high variability of symbol appearance. Rotation, partial occlusions, elastic deformations, or...
Variations in illumination can have a dramatic effect on the appearance of an object in an image. In this paper, we propose how to deal with illumination variations in eigenspace ...
Classification of images in many category datasets has
rapidly improved in recent years. However, systems that
perform well on particular datasets typically have one or
more lim...
Action recognition is a popular and important research topic in computer vision. However, it is challenging when facing viewpoint variance. So far, most researches in action recogn...