Automatic VOE Techniques using CRF and Visual Saliency Method

Object extraction from active videos is by and large an unexplored area in computer vision and pattern recognition. We have designed several analysis algorithms for video object detection and segmentation in the general framework of multimedia content analysis. The algorithms are developed in two directions, i. e. , fully automatic algorithms and semi-automatic algorithms. In this project we can propose a new method of video object detection to automatically extract the objects from acquired videos. The existing video object detection techniques operate under homogenous object motions. In contrast, our proposed video object extraction approach effectively detects the object under motion sequences and robust a Hidden Markov Model (HMM), Support Vector Machine (SVM) and Gaussian Mixture Model is applied to automatically determine the label of each pixel based on the observed models. Then we deal with multiple object instances and this system does not require the previous knowledge of the object.