Here, we present a constrained object recognition task that has been robustly solved largely with simple machine learning methods, using a small corpus of about 100 images taken u...
Human action video sequences can be considered as nonlinear dynamic shape manifolds in the space of image frames. In this paper, we address learning and classifying human actions ...
– This paper addresses the exploitation of massive numbers of image-derived change detections. We use the term “change analysis” to emphasize the intelligence value obtained ...
Natural scene images coming usually from lowresolution sensors in embedded context suffer from low text recognition results. Due to several types of degradations, existing algorit...
An offline recognition system for Arabic handwritten words is presented. The recognition system is based on a semi-continuous 1-dimensional HMM. From each binary word image norma...