Temporal Conditional Random Fields: A conditional state space predictor for visual tracking

摘要

We present a modified Temporal Conditional Random Fields framework for modeling and predicting object motion. To facilitate such a powerful graphical model with prediction and come up with a CRF-based predictor, we propose a set of new temporal relations for object tracking, with feature functions such as optical flow (calculated among consequent frames). We evaluate our proposed Temporal Conditional Random Field method with real and synthetic data sequences and will show that the TCRF prediction is nearly equivalent with result of template matching. Experimental results show that our proposed method estimates future target state with zero error until target dynamic changes. Our proposed modified CRF method with simple and easy to implement feature functions, can learn any target dynamic, thus, it can predict next state of target with zero error.