Learning a family of detectors via multiplicative kernels.
IEEE Trans. Pattern Anal. Mach. Intell.(2011)
摘要
Object detection is challenging when the object class exhibits large within-class variations. In this work, we show that foreground-background classification (detection) and within-class classification of the foreground class (pose estimation) can be jointly learned in a multiplicative form of two kernel functions. Model training is accomplished via standard SVM learning. When the foreground object masks are provided in training, the detectors can also produce object segmentations. A tracking-by-detection framework to recover foreground state in video sequences is also proposed with our model. The advantages of our method are demonstrated on tasks of object detection, view angle estimation, and tracking. Our approach compares favorably to existing methods on hand and vehicle detection tasks. Quantitative tracking results are given on sequences of moving vehicles and human faces.
更多查看译文
关键词
multiplicative kernels,quantitative tracking,video signal processing,kernel functions,foreground-background classification,angle estimation,large within-class variation,vehicle detection,video sequences,object class,foreground object mask,learning (artificial intelligence),image segmentation,vehicle detection task,object segmentation,object segmentations,pose estimation,feature extraction,image classification,kernel methods.,object tracking,image sequences,object detection,family of detectors,object recognition,computer vision,foreground background classification,model training,tracking-by-detection framework,foreground class,standard svm learning,foreground state,support vector machines,multiplicative kernel,within class classification,markov chains,kernel,foreground background,kernel methods,detectors,motion,kernel method,kernel function,learning artificial intelligence,indexing terms,algorithms,computer simulation,artificial intelligence,support vector machine
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络