Monocular Multiview Object Tracking With 3D Aspect Parts

The Monocular Multiview Object Tracking with 3D Aspect Parts Dataset is used to study the problem of tracking objects under viewpoint changes, including viewpoint changes, car tracking standard sequences, etc.

The publisher of this dataset proposed a new method, which is mainly used to track objects, estimate continuous poses and part positions under severe viewpoint changes. In order to deal with topological appearance changes caused by viewpoint illusion, 3D aspect parts are used to represent objects, and the relationship between viewpoints and 3D aspect parts is simulated in a component-based particle filtering framework.

In addition, the dataset demonstrates instance-level online learning of part appearance and examples integrated into the model, making it more robust in difficult scenes with occlusions.

This dataset was released by the Computational Vision and Geometry Laboratory of Stanford University in 2014. The related paper is "Monocular Multiview Object Tracking with 3D Aspect Parts".